Running Head: TRENDS IN ACADEMIC PERFORMANCE Uneven Progress: Recent Trends in Academic Performance Among U.S. School Districts

We use data from the Stanford Education Data Archive to describe district-level trends in average academic achievement between 2009 and 2018. Though on average school districts’ test scores improved by about 0.003 SDs per year, there is significant variation among districts. Additionally, we find that average test score disparities between non-poor and poor students are growing, those between White and Black students are stagnating, and those between White and Hispanic students are shrinking. We find no evidence of achievement-equity synergies or negative tradeoffs: improvements in overall achievement are uncorrelated with trends in achievement gaps. Finally, we find that the strongest predictors of achievement gap trends are the levels and trends in within-district racial and economic segregation and inequality. TRENDS IN ACADEMIC PERFORMANCE 3 Average standardized test scores in the United States have risen since the 1970s, particularly among elementary and middle school students (NCES, 2013). The average 4th grader in 2019, for example, had a mathematics test score that was nearly a standard deviation higher than those of 4th graders in her parents’ generation in the 1990s (NCES, 2020). These increases reflect improved educational opportunities: they imply that children growing up today have, on average, more resources and opportunities—in their homes, neighborhoods, preschools, and elementary and middle schools—to learn the math and reading skills measured by standardized tests than did children 50 years ago. The increase in average test scores is evident for all racial/ethnic groups, though the increases have been larger among Black and Hispanic students than among White students. As a result, the national White-Black and White-Hispanic achievement gaps have narrowed substantially in the last 50 years as well (Reardon et al., 2015). In contrast, the achievement gap between nonpoor and poor students (as measured by free/reduced-price lunch eligibility) has been relatively stable for several decades (NCES, 2020), while the gap between affluent and very poor students (those at the 90th and 10th percentiles of the income distribution, respectively) has widened substantially (Reardon, 2011; but see also Hashim et al., 2020; Reardon, 2020). On the one hand, the pattern of rising average test scores and narrowing racial/ethnic achievement gaps indicates that increases in overall educational opportunities have been accompanied—at least over the long term—by growing racial equity of opportunities. On the other hand, the same increases in overall educational opportunity have been accompanied by stable or widening economic achievement gaps; economic inequality of educational opportunity has not improved along with overall educational outcomes. TRENDS IN ACADEMIC PERFORMANCE 4 These national trends (as well as state-level trends in the last two decades) in test scores are well-documented by the National Assessment of Educational Progress (NAEP). But what has happened at a more local scale? The U.S. educational system is highly decentralized, with over 13,000 public school districts, each of which has considerable control over staffing, curriculum, instruction, and budget allocation decisions. To what extent are the national trends in average test scores and test score gaps common among school districts? How much do these trends vary and covary? What local and schooling characteristics are associated with local trends in achievement patterns? Are there many districts where scores are improving while gaps are narrowing, or does improvement generally come with increased inequality? Our goal in this paper is to answer these questions, providing a detailed descriptive account of trends in achievement patterns among all school districts in the U.S. Our work here extends the recent work of Atteberry et al (in press) on this topic. In this article, we first measure recent trends in students’ academic performance in every public school district in the United States using ten years of data (2009 to 2018) on thirdthrough eighth-grade math and reading test scores. We show that we are able to measure district-level trends with high reliability. We then describe how these trends vary among school districts and how they differ between socioeconomic and racial/ethnic student subgroups within school districts. Third, we estimate the correlation between trends in overall performance and trends in racial and economic achievement gaps, in order to determine whether increasing overall performance and reducing achievement gaps are synergistic or conflicting processes. Finally, we estimate the correlation of trends in average performance and in racial and economic achievement gaps with local demographic changes and school characteristics. This exploratory TRENDS IN ACADEMIC PERFORMANCE 5 analysis helps identify potential factors to examine further as levers for increasing opportunity and student achievement. We find that, in the average district, test scores changed very little (increasing by only 0.003 standard deviations (SDs) per year) from 2009 to 2018 (less than one-tenth of a grade level over a decade). The trends vary considerably among districts, however: test scores improved by more than 0.025 SDs per year (roughly three-quarters of a grade level over the study period) in the fastest-improving one-sixth of districts, and declined by 0.020 SDs or more per year (more than half a grade level over the study period) in the one-sixth with the fastest-decreasing performance. Changes in districts’ socioeconomic and racial/ethnic composition explain very little of the variation in academic performance trends. Rather, we find that the single strongest predictor of improvement is average test scores at the start of the period. Districts with initially high average levels of educational opportunity experienced the greatest improvement over the study period. As a result, average achievement varied more across districts in later cohorts than earlier cohorts. We also find that district-level trends in academic performance vary significantly by subgroup. In the average district, scores for both poor and nonpoor students improved, on average, over the study period, but the improvement was greater for nonpoor students, so that the average nonpoor-poor achievement gap widened by 0.005 SDs/year from 2009 to 2018 (an increase of roughly 10% over a decade). The White-Black gap also widened very slightly (by 0.002 SDs/year) in the average district, while the White-Hispanic gap narrowed (by 0.005 SDs/year). The trends in all three gaps varied substantially among districts, however. Moreover, the strongest predictors of increasing achievement gaps are measures of economic inequality and TRENDS IN ACADEMIC PERFORMANCE 6 segregation: achievement gaps have grown most rapidly, on average, in school districts with high and increasing levels of social inequality and school segregation. Our findings, in conjunction with other research, suggest that reducing within-district segregation and inequality may lead to greater equality of educational opportunity.