Recent years have seen a dramatic expansion in the number of states and school districts requiring students to demonstrate basic proficiency on standardized tests in certain grades and subjects in order to be default promoted to the next grade level. Such test-based promotion policies aim to end the common practice of social promotion, or advancing students to the next grade level even if they are academically unprepared. Test-based promotion policies have been in effect for several years in Florida, New York City, and Chicago; they were recently adopted by Oklahoma, Arizona, Colorado, and Indiana; and other states including Ohio, North Carolina, Iowa, New Mexico, and Tennessee are considering similar legislation. Each of these policies focuses at least in part on retaining students who have not demonstrated basic reading proficiency in third grade, often considered a key milestone in students’ academic development.

Until recently, the near-consensus opinion among education researchers was that test-based promotion policies were wrongheaded because retention is associated with lower student achievement and an increased probability of dropping out prior to graduating (see, e.g., Jimerson 2002). Because retention decisions typically reflect unobserved student characteristics, however, this association may be due to selection bias.

More recent research has exploited test-based promotion policies to study the causal impact of grade retention itself or combined with remedial interventions on student test scores, and tends to find more positive results for the effect of retention on student test score performance than did prior research (Jacob and Lefgren 2004, Winters and Greene 2012, and Schwerdt and West 2013).

Thus far, however, there is little evidence on the effect of test-based promotion policies on high school course choice and graduation. We have no such evidence on the effects of policies that retain low-performing students in the third grade, the most common policy design. Analyzing the effects of such test-based promotion policies on retained students’ chances of completing high school and college readiness is essential to determine their overall impact on students and society.

We use a fuzzy regression discontinuity design to estimate the causal impact of retention under Florida's third grade test-based promotion policy on students’ course choices in high school and their probability of on-time graduation. Florida’s policy is of particular interest because it is the longest-running statewide test-based promotion policy and has served as a model for policies adopted elsewhere. Longitudinal administrative data enable us to follow the universe of public school students in the state from the entering third grade cohort of 2002-03 -- the first cohort subject to the policy -- through the end of the 2012-13 school year, when students retained under the policy were expected to have completed high school.

Our results will provide valuable new information about the consequences of third grade test-based promotion policies to inform the discussion about such programs across the nation. The pace with which these policies are being adopted makes evidence on their impact on educational attainment timely.