The Achievement Gap, Revisited: An Empirical Assessment of What We Can Learn from East Asian Education

Abstract

International mathematics assessments have established students in East Asia as among the best in the world and their U.S. counterparts as mediocre. What is not clear is why this “achievement gap” exists. The last major study to address this question, Stevenson and Stigler’s (1992) The Learning Gap, was published prior to empirical and methodological advances in international comparative research on education. Prevailing wisdom points to unverified differences in cultural beliefs, which often leads to defeatist conclusions. This dissertation offers a fresh perspective by applying sociological theory and methods to the issue. Specifically, I rely on underutilized data from the 2003 and 2007 Trends in International Mathematics and Science Study (TIMSS) of fourth graders to compare educational systems across three major factors that influence math achievement: curriculum, teachers and parents. My main empirical findings are that there is greater uniformity of math instruction across classrooms in the participating East Asian nations of Hong Kong, Japan, Singapore and Taiwan than in the U.S. and that, among all participating educational systems, average achievement tends to be higher in those with greater uniformity of instruction. The implication is that the institutional arrangements that allow for less uniformity of instruction across classrooms in the U.S. might be partially responsible for the gap. Cross-regional differences in teacher effectiveness might also account for part of the gap, as three-level, hierarchical linear models of achievement in each nation indicate that U.S. math teachers are less effective than their East Asian counterparts—even after the quantity of instruction provided is taken into account. The main theoretical contribution is an alternative explanation for the apparent cross-regional disparity in the proportion of involved parents, which highlights how schools can make a difference in whether or not parents become involved. Such an approach promises a way out of the dead-end reached by previous theorists. However, this dissertation also draws attention to the limitations of the existing data. At present, there is not enough information available to substantiate the policy recommendations made in previous studies. As such, a central aim of this dissertation is to put research onto sounder methodological footing.