Panel Paper: Early Impacts From a Cluster Randomized Trial of “Promise” College Financial Aid Programs

Saturday, November 10, 2012 : 11:15 AM
International E (Sheraton Baltimore City Center Hotel)

*Names in bold indicate Presenter

Douglas Harris, University of Wisconsin at Madison

College financial aid programs suffer from many flaws, including that the amount of aid they will receive is uncertain until just before they reach college age.  A simple and essentially costless policy change would be to make aid commitments much earlier, while students are in middle and high school when it still has a chance to convince them that college is an affordable option and thus induce them to take more academically challenging courses and learn about colleges and their application processes.  On the other hand, if students’ high school decisions are unaffected by expectations about uncertain aid, then the timing of the aid may be irrelevant. This study tests these competing hypotheses with a cluster randomized trial, funded by $30 million from a non-profit organization.  In November, 2011 all first-time ninth graders in 18 Milwaukee high schools were selected to receive a $12,000 college scholarship and the remaining 18 schools serve as controls.  Students receive letters three times per year reminding them of the scholarship and indicating whether they are on-track to meet the program’s two main requirements: achieving a 2.5 GPA and attending class 90% of the time (cumulative across years), requirements that mirror the Pittsburgh Promise.  To increase statistical power, the 18 schools were selected via paired randomization and baseline equivalence tests confirm that the two groups are essentially identical at baseline.  We recently received extensive data administrative and survey data from the Milwaukee Public Schools for essentially all students in the 36 schools.  In addition to posing some alternatives to standard tests of baseline equivalence, our presentation will report the first-year impacts of the scholarship experiment on grades, attendance, and disciplinary incidents, as well as how these effects vary based on students’ pre-treatment proximity to the GPA and attendance cut-offs.