Research Papers

Can School Accountability Pressure Help Disadvantaged Students Catch up? Evidence From No Child Left Behind (Job Market Paper)

Though the central goal of No Child Left Behind (NCLB) was to close achievement gaps, whether school accountability pressure under NCLB has succeeded in helping disadvantaged students catch up to their more advantaged peers is surprisingly unclear. In this paper, I investigate this fundamental research question and explore potential underlying mechanisms to inform the design of future school accountability policies. I use administrative data from North Carolina and adopt a difference-in-differences strategy. My primary estimation method is proposed by Gardner (2022), and I examine the robustness of the results with other alternative methods. The main threat to the identification is the validity of the parallel-trends assumption; with event study methods, I directly check disparities in pre-trends and find no or minor differences. I find that school accountability pressure in math, measured by a school failure in math, produces significantly larger effects on math z-scores among the traditionally low-performing student subgroups: black, Hispanic, economically disadvantaged, and students with disabilities. Furthermore, I show these larger effects in math among disadvantaged groups are driven by the larger effects among low-performing students and the fact that significantly more students in these disadvantaged groups are low-performing. I find smaller effects of reading accountability pressure on reading z-scores, and the largest effect appears among Hispanic students.

Research in Progress

Gender Difference in the Relationship Between Obesity and Academic Performance: Association Versus Causal Impact (with Stephanie Coffey and Amy Ellen Schwartz)

The existing literature using survey data finds conflicting evidence on the association between obesity and academic performance among boys and girls. In this paper, we leverage unique student-level longitudinal data from New York City (NYC) to examine the gender difference in the relationship between obesity and student performance. Our findings help explain the mixed results documented in the existing literature. Specifically, we find a larger raw association between obesity and academic performance among girls than boys. Obese girls perform 0.103 (0.133) standard deviations (sd.) lower in math (ELA) than normal weight girls, while obese boys score 0.039 (0.077) sd. lower in math (ELA). However, after accounting for student demographics and time-invariant characteristics, we find a larger association among boys than girls. Obese girls perform 0.009 (0.012) sd. lower in math (ELA), while obese boys score 0.021 (0.016) sd. lower in math (ELA). With a regression discontinuity design, we find some evidence of a causal impact of being classified as obese on student test scores for boys but not girls. Specifically, we find that being classified as obese leads to a decrease of 0.014 sd. in math and 0.013 sd. in reading among boys. Our results suggest a negative association between obesity and academic performance among girls but a deleterious causal effect – albeit a small effect – among boys.

School Accountability and Student Composition in the 165 Most Selective Public High Schools

Though selective secondary schools provide excellent educational opportunities to high-achieving students, students from disadvantaged populations are underrepresented in those institutions. One plausible reason is the challenges disadvantaged students encounter in accessing and preparing for the entrance exams. With the introduction of state accountability policies in the 1990s, some elite schools were able to incorporate student performance in state-wide standardized tests into their admission criteria, and schools were motivated to improve the performance of disadvantaged students. In this paper, I investigate how school accountability affects student composition in the 165 most selective high schools, those classified via a rigorous procedure by Finn and Hockett (2012). I use the Common Core Data from 1987-2002, and I compare elite public high schools with regular public high schools in the same districts before and after school accountability policies were implemented in their states. Preliminary results find after school accountability, proportions of black, Hispanic, and economically disadvantaged students increase in both regular and selective high schools, while the proportion of white students drops.

Extra Pressure, Extra improvement? Effect of a Subgroup Failure Under No Child Left Behind

A crucial feature of No Child Left Behind is that schools were held accountable for the performance of all students and up to 9 student subgroups. This subgroup rule was intended to draw schools’ attention to low-performing subgroups. However, it remains unclear whether a subgroup failure produces additional effects for students in the group after accounting for the effect of a school failure. In this paper, I use student-level data from North Carolina public schools from 2003-2006 to study the extra effects of a subgroup failure. I adopt two identification strategies: a difference-in-differences strategy and a regression discontinuity design. To disentangle the effect of a subgroup failure from the effect of a school failure, I only include student-school-year observations that a school failed, which means all the observations were under the treatment of a school failure.