Co-Principal Investigator: Eric Bettinger (Stanford University)

Purpose: In this project, researchers investigate whether an intervention that provides financial aid information to college students increases receipt of financial aid, enrollment on a full-time (versus part-time) basis, persistence from semester to semester, and degree completion. If proven beneficial, the intervention could be widely replicated based on information that is already available in federal and institutional databases. The goal of the intervention is to alter students' perceptions of the costs and benefits of college, and in particular, reducing the short-term costs of attendance by increasing financial aid receipt. Results of this intervention will interest policymakers at the institution and system levels.

Project Activities: In the first stage of the project, researchers will collect the administrative data necessary to personalize the information that they will send to individual students, and will also finalize the content and formatting of information that they will send to all students. In the middle stage of the project, researchers will carry out the intervention by providing a series of emails and letters to students aimed at informing them about requirements for receiving financial aid, and to part-time students informing them of the economic benefits of attending college on a full-time basis. They will conduct a follow-up survey to capture students' experiences with the intervention, and will also track students' progress through postsecondary education and their completion of degrees. In the final stage of the project, researchers will analyze the impacts of the intervention on persistence and degree completion, the factors that make the intervention more or less effective, and the student groups for whom the intervention appears most helpful.

Products: Researchers will generate evidence of the efficacy of their informational intervention for U.S. college students at open- and broad-access institutions, and will produce a variety of non-technical reports and other communications to convey their findings to policymakers and administrators. The research team will also produce peer-reviewed publications.

Structured Abstract

Setting: Students will be enrolled at open- and broad-access institutions spread out across the United States for this project.

Sample: Approximately 12,000 first- and second-year students attending open- and broad-access postsecondary institutions will participate in the study. The sample will come from students participating in the National Postsecondary Student Aid Study of 2016 (NPSAS:16) who were interviewed at the time when they were recruited for NPSAS:16, and who agreed to consider participation in additional data collections associated with NPSAS:16. The sample will be representative of first- and second-year students attending open- and broad-access postsecondary institutions in the United States.

Intervention: The intervention will consist of mailed letters and emails that contain guidance on how to obtain financial aid by completing the Free Application for Federal Student Aid (FAFSA). Yearly completion of the FAFSA is crucial to receiving financial aid because renewal of federal aid from year to year requires FAFSA completion, and most other forms of aid also require FAFSA completion. Researchers will randomize key features of the intervention including whether the letters or emails contain messages with negative, neutral, or positive framing, and whether included information describes the relationship between enrollment intensity and available financial aid. Researchers will also personalize the messaging so that it addresses tasks of importance to students at specific institutions and to specific students.

Research Design and Methods: Researchers will randomize students to the treatment and control conditions. Among the group of 6000 full-time students there will be four equally-sized groups: control; and neutral, negative, and positive framing. Among part-time students, there will be three groups: control, students receiving FAFSA application info, and students receiving this information plus information about the benefits of increased course taking.

Control Condition: Students in the control condition will receive the business-as-usual financial aid information and reminders provided by their institutions.

Key Measures: During the first year of the study, researchers will measure enrollment intensity (full- or part-time), credits attempted and completed, and term-to-term persistence using transcript data included in NPSAS:16. After the first year, researchers will track enrollment intensity, term-to-term persistence, and degree completion using data from the National Student Clearinghouse. Potential moderators of the intervention such as a student's race, gender, family income, and first-generation status will come from NPSAS:16. In line with prior research, mediators will include filing of the FAFSA and subsequent receipt of financial aid, coded from data within the National Student Loan Data System (NSLDS). The research team will track other potential mediators (e.g. students' engagement in college, students' understanding of the financial aid process, and institutional supportiveness) through baseline and follow-up surveys administered by the research team. Researchers will also measure implementation fidelity through items on the follow-up survey that ask students about their engagement with the intervention.

Data Analytic Strategy: Researchers will assess baseline equivalence across the treatment and control groups. If any significant differences are found, researchers will include the specific covariates necessary to balance the two groups in all subsequent analyses. To assess the intervention in relation to their full set of outcomes of interest, the researchers will use linear regression models with covariates to increase the precision of the estimated treatment impacts. The researchers will test for clustering of correlations among students within institutions as well as large and non-normal variation in unexplained variance among individual students, and in the event that either of these problems is present, will employ multi-level models and/or robust standard errors, respectively.