Could teacher grading practices account for unexplained variation in school achievements?

Abstract

In the economics of schooling literature, input substitutions have been identified as the major difficulty in assessing school and teacher effectiveness. In this paper we examine whether a teacher's grading practice can improve students' academic achievements by reducing these input substitutions. We use the teacher-student interaction model of Correa and Gruver (1987) to enhance the specification of the traditional production functions of students' academic achievements by introducing a teacher grading parameter in these functions. Under specific conditions, Correa and Gruver show that a student may respond either positively or negatively to the teacher's greater effort and/or harder grading, depending upon the student's preferences; thus the issue of an efficient grading practice is, as usual in the economics of education research, an empirical issue. To implement the econometric specification of the model, we define a latent grading variable from an error components specification on class grade regressions. This variable is used to explain the students' grade specific standardized test results for a sample of first and fourth-graders of Montreal francophone public elementary schools.