Patterns of Change in U.S. Gender Achievement Gaps During Elementary and Middle School

• In Event: Patterns and Correlates of Gender Achievement Gaps in the United States

Abstract

This paper seeks to understand the systematic patterns in how gender achievement gaps grow, shrink or stay constant as students move through elementary school and middle school. The focus of this paper is not only to capture these trends for average gender achievement gaps, but also for gaps in the tails of the achievement distribution as those have been demonstrated critical areas of analysis by prior work in the field. Research on gender achievement gaps shows they exist across the distribution, exist as early as Kindergarten, and persist through 8th grade. However, past research does not explicitly model the growth of gender achievement gaps across grade-levels to characterize key trends.

To conduct this analysis, this study leverages a longitudinal dataset provided by the Northwest Evaluation Association (NWEA). The NWEA test, the Measures of Academic Progress (MAP) assessment, is a low-stakes formative assessment designed to help teachers better evaluate and assist their students throughout the school year. The data include fall and spring test scores for nine grades (Kindergarten through 8th grade) over nine school years (2005 through 2013) with test records for approximately 7 million students, 15,000 schools, and 3,700 districts in the dataset from all U.S. states.

I adopt metric-free measures of the mean and tail gender achievement gap, the V-Statistic (Ho, 2009; Ho & Haertel, 2006; Ho & Reardon, 2012) and the Proportion-Adjusted Relative Difference (Robinson & Lubienski, 2011), respectively. The use of metric free measures enables the comparison of gender achievement gaps across grades without concerns about equal interval scales or comparable test metrics. To analyze the patterns in the gaps across grades, I adopt the “change in gap” framework, modeling changes in the subject-by-cohort-by-grade-by-term gaps using Hierarchical Linear Modeling (HLM) with multiple functional forms for the grade variable. Investigating multiple functional forms stems from trying to narrow the hypothesized mechanisms by understanding if gaps consistently increase or decrease over grades (linear trend) or if the changes are more sporadic (nonparametric trend).

I find that gender achievement gaps in both mathematics and reading change meaningfully as students progress through grades. Specifically, changes in the gaps are best captured using a nonparametric trend. The linear trend masks the period of increase followed by a period of decrease in the gaps at different grades. In both subjects, the trend in the average gaps that emerges is that gaps change in favor of males in elementary school (growing the mathematics gap, shrinking the reading gap), and in favor of females in middle school (shrinking the mathematics gap, growing the reading gap). These trends hold for the upper and lower tail gaps, as well, with the exception of the lower tail in reading where male students continuously fall behind.

This paper adds to the current literature through systematically analyzing and characterizing the changes in the gaps as students move through elementary and middle school. This analysis further provides intuition as to the potential mechanisms driving these systematic changes, sparking critical lines of future research.

Author

• Erin Michelle Fahle, Stanford University