Purpose: A growing number of states require students to pass exit exams in order to graduate from high school. However, a large and vocal group of testing opponents worries that exit exams will harm students, especially disadvantaged students who tend to do poorly on standardized tests. Critics argue that exit exams lower graduation rates by denying diplomas to students who complete all other requirements for a degree, and by inducing some students who fail the exam initially to drop out rather than try again. Previous research by one of the Principal Investigators for this project examines the relation between exit exam performance and other student outcomes using data from the state of Texas. This research will use data from Florida students to: (1) assess the generalizability of the Texas results; (2) see if specific features of a testing policy (such as the number of retest attempts) matter; (3) test whether the relation between exit exam performance and other student outcomes is moderated by student characteristics; and (4) conduct a longer follow-up than was possible in Texas to see if exit exam performance is related to long-term economic outcomes.

Project Activities: The researchers will use standard statistical techniques to estimate the discontinuity, or "jump," in each of the key outcomes at the exit exam passing cutoff. The magnitude of the discontinuity will be interpreted as the impact of passing status. Because students can retake the test in both Florida and Texas if they initially fail, the researchers will analyze the impact of passing status separately by test administration. This will provide evidence on whether the impact of failing the exit exam early in high school differs from that of failing at the end of senior year. The researchers will then compare results obtained from Florida to the earlier findings from Texas. The testing systems in the two states differ in important ways, and this comparison will shed light on whether policy design features matter. They will also compare the results of academic cohorts subject to different testing policies (such as the number of retest opportunities) to examine the importance of specific policy attributes. Finally, they will conduct a set of analyses separately by student subgroups to see if exit exam performance has an especially strong impact on certain types of students.

Products: Products from this project include a better understanding of the impact of exit exams on high school completion, and published papers.

Purpose: A growing number of states require students to pass exit exams in order to graduate from high school. However, a large and vocal group of testing opponents worries that exit exams will harm students, especially disadvantaged students who tend to do poorly on standardized tests. Critics argue that exit exams lower graduation rates by denying diplomas to students who complete all other requirements for a degree, and by inducing some students who fail the exam initially to drop out rather than try again. Previous research by one of the Principal Investigators for this project examines the impact of exit exam performance on student outcomes using data from the state of Texas. This research will use data from Florida students to: (1) assess the generalizability of the Texas results; (2) see if specific features of a testing policy (such as the number of retest attempts) matter; (3) test whether the impact varies by student characteristics; and (4) conduct a longer follow-up than was possible in Texas to see if exit exam performance affects long-term economic status.

Setting: The schools are in Texas and Florida.

Population: The data will be obtained through administrative records that contain information on all students enrolled in the public high schools of Florida and Texas. All students subject to the exit exam requirement will be studied. About 105,000 students belong to the test-taking cohorts for the period covered in this project. In Texas, the researchers will study students who initially took the exit exam between 1991 and 1995. In Florida, the researchers will use information on students in the 1991-1999 high school graduation classes.

Intervention: The data from Texas comes from the Texas Schools Microdata Panel, which is a collection of administrative records from various Texas state agencies. Individual student records in this database can be linked over time, making it possible to follow students as they progress through high school and then enter college or the workforce. The Florida data comes from the Florida Education and Training Placement Information Program, a database that provides information on employment, continuing postsecondary education, military, public assistance participation, and incarceration. As in Texas, the researchers will be able to link the records of high school exit exam takers to information on graduation, enrollment in college, and labor market earnings.

Research Design and Methods: The researchers will use a quasi-experimental regression discontinuity design. The basic idea underlying this approach is to compare the outcomes of students who barely fail and barely pass the exam. This comparison is useful because the differences between the average passer and failer are likely to disappear when one focuses on students with test scores in an increasingly narrow range. Thus, differences in the outcomes of barely passers and barely failers can plausibly be attributed to passing status per se, rather than to some unobserved factor. The key assumption for this methodology is that any determinant of the outcome being studied (other than passing status) trend smoothly at the passing cutoff.

Key Measures: The researchers will analyze high school outcomes including dropout behavior and graduation as well as post-high school outcomes such as college attainment and labor market earnings.

Data Analytic Strategy: The researchers will use standard statistical techniques to estimate the discontinuity, or "jump," in each of the key outcomes at the exit exam passing cutoff. The magnitude of the discontinuity will be interpreted as the impact of passing status.

Products and Publications

Journal article, monograph, or newsletter

Clark, C. and Martorell, P. (2014). The Signaling Value of a High School Diploma. Journal of Political Economy, 122 (2): 282–318.

Clark, D., and See, E. (2011). The Impact of Tougher Education Standards: Evidence from Florida. Economics of Education Review, 30 (6): 1123–1135.

** This project was submitted to and funded under Middle and High School Reform in FY 2006.