Association for Public Policy Analysis & Management

Many high-achieving low-income students do not enroll in or persist through college (Hoxby & Avery, 2013). Strategies to dismantle informational or administrative hurdles to the application process, such as in-person application assistance or text messages connecting students to counselors have shown substantial impacts (Carrell and Sacerdote, 2018; Bettinger et al., 2013; Castleman & Page, 2016). However, these approaches face possible scale-up challenges and are expensive relative to lighter-touch information interventions. While the evidence is mixed on the effectiveness of such interventions, Barr and Turner (2018) and Hoxby and Turner (2013) show promising impacts, but focus on subpopulations that may not be representative of the typical high-achieving low-income student. Policy-makers are still in search of the cheapest, scalable intervention to induce poor, moderately high-achieving students not already on the college path into appropriately selective colleges.

During fall 2014, I mailed letters from the Michigan Department of Education to a randomly assigned half of the nearly 50,000 Michigan seniors who scored at least a 20, the median score, on the ACT, which all students take in school. The letter contained a link to a publicly available website containing information about the college and financial aid application processes, as well as an individual-specific password allowing me to track who navigates to the website and their browsing behavior. I matched the sample to the National Student Clearinghouse to examine whether letter receipt influenced college enrollment and persistence.

Approximately ten percent of treated students navigated to the website, though this varied dramatically by student characteristics. Poor, high scoring students and black, high scoring students had the highest take up rates (16.9% and 17.7%, respectively), suggesting that these groups of students are most in need of information and support in the college application process. I find that tiny changes to letter content had dramatic effects on take-up. For example, including the phrase “Learn how to make college affordable” produced a take-up rate 18% higher than including the phrase “Learn which college is right for you,” and 37% higher than the phrase “Learn how to apply to college”, with take-up rates of 12.0%, 10.2%, and 8.7%, respectively. These differences represent a revealed-preference approach to determining the barriers that students face when applying to college. Consistent with these results, students are more likely to navigate to pages on the website related to college affordability.

Finally, I find that poor students who were mailed the letter were 1.5 percentage points, or 2 percent, more likely to enroll in college, driven by increases at four-year colleges. However, these effects did not translate into increased enrollment beyond the first year of college. The probability of enrolling and persisting to the second year of college increased by a statistically insignificant 0.5 percentage points, and by only 0.2 percentage points for enrolling and persisting to year three. These results emphasize the need for policy-makers to focus efforts on increasing persistence rates through college, and highlight for researchers the critical importance of examining whether students “nudged” into college drop-out or persist toward a degree.