- Benjamin L. Castleman
Search EdWorkingPapers by author, title, or keywords.
Benjamin L. Castleman
In-person college advising programs generate large improvements in college persistence and success for low-income students but face numerous barriers to scale. Remote advising models offer a promising strategy to address informational and assistance barriers facing the substantial majority of low-income students who do not have access to community-based advising, yet the existing evidence base on the efficacy of remote advising is limited. We present a comprehensive, multi-cohort experimental evaluation of CollegePoint, a national remote college advising program for high-achieving low- and moderate-income students. Students assigned to CollegePoint are modestly more likely (1.3 percentage points) to attend higher-quality institutions. Results from mechanism experiments we conducted within CollegePoint indicate that moderate changes to the program model, such as a longer duration of advising and modest expansions of the pool of students academically eligible to participate, do not lead to larger program effects. We also capitalize on across-cohort variation in whether students were affected by COVID-19 to investigate whether social distancing required by the pandemic increased the value of remote advising. CollegePoint increased attendance at higher-quality institutions by 3.2 percentage points for the COVID-19-affected cohort. Acknowledgements.
Colleges have increasingly turned to predictive analytics to target at-risk students for additional support. Most of the predictive analytic applications in higher education are proprietary, with private companies offering little transparency about their underlying models. We address this lack of transparency by systematically comparing two important dimensions: (1) different approaches to sample and variable construction and how these affect model accuracy; and (2) how the selection of predictive modeling approaches, ranging from methods many institutional researchers would be familiar with to more complex machine learning methods, impacts model performance and the stability of predicted scores. The relative ranking of students’ predicted probability of completing college varies substantially across modeling approaches. While we observe substantial gains in performance from models trained on a sample structured to represent the typical enrollment spells of students and with a robust set of predictors, we observe similar performance between the simplest and most complex models.
Growing experimental evidence demonstrates that low-touch informational, nudge, and virtual advising interventions are ineffective at improving postsecondary educational outcomes for economically-disadvantaged students at scale. Intensive in-person college advising programs are a considerably higher-touch and more resource intensive strategy; some programs provide students with dozen of hours of individualized assistance starting in high school and continuing through college, and can cost thousands of dollars per student served. Despite the magnitude of this investment, causal evidence on these programs' impact is quite limited, particularly for programs that serve Hispanic students, the fastest growing segment of U.S. college enrollees. We contribute new evidence on the impact of intensive college advising programs through a multi-cohort RCT of College Forward, which provides individualized advising from junior year of high school through college for a majority Hispanic student population in Texas. College Forward leads to a 7.5 percentage point increase in enrollment in college, driven entirely by increased enrollment at four-year universities. Students who receive College Forward advising are nearly 12 percentage points more likely to persist to their third year of college. While more costly and harder to scale than low-touch interventions, back of the envelope calculations suggest that the benefit from increased college graduation likely induced by the program outweighs operating costs in less than two years following college completion.
The Post-9/11 GI Bill allows service members to transfer generous education benefits to a dependent. We run a large scale experiment that encourages service members to consider the transfer option among a population that includes individuals for whom the transfer benefits are clear and individuals for whom the net-benefits are significantly more ambiguous. We find no impact of a one-time email about benefits transfer among service members for whom we predict considerable ambiguity in the action, but sizeable impacts among service members for whom education benefits transfer is far less ambiguous. Our work contributes to the nascent literature investigating conditions when low-touch nudges at scale may be effective. JEL Classification: D15, D91, H52, I24
Do nudge interventions that have generated positive impacts at a local level maintain efficacy when scaled state or nationwide? What specific mechanisms explain the positive impacts of promising smaller-scale nudges? We investigate, through two randomized controlled trials, the impact of a national and state-level campaign to encourage students to apply for financial aid for college. The campaigns collectively reached over 800,000 students, with multiple treatment arms to investigate different potential mechanisms. We find no impacts on financial aid receipt or college enrollment overall or for any student subgroups. We find no evidence that different approaches to message framing, delivery, or timing, or access to one-on-one advising affected campaign efficacy. We discuss why nudge strategies that work locally may be hard to scale effectively.
Student loan borrowing for higher education has emerged as a top policy concern. Policy makers at the institutional, state, and federal levels have pursued a variety of strategies to inform students about loan origination processes and how much a student has cumulatively borrowed, and to provide students with greater access to loan counseling. We conducted an experiment to evaluate the impact of an outreach campaign that prompted loan applicants at a large community college to make informed and active borrowing decisions and that offered them access to remote, one-onone assistance from a loan counselor. The intervention led students to reduce their unsubsidized loan borrowing by 7 percent, resulted in worse academic performance, and increased the likelihood of loan default during the three years after the intervention occurred. Our results suggest policy makers and higher education leaders should carefully examine the potential unintended consequences of efforts to reduce student borrowing, particularly in light of growing evidence regarding the counter-intuitive positive relationship between reduced borrowing levels and worse student academic and financial outcomes.
Despite broad public interest in Veterans' education, there is relatively little evidence documenting the postsecondary trajectories of military service members after they return to civilian life. In the current report we investigate how U.S. Army service member college enrollment and progression trends compare to a similar population of civilians, using Army administrative personnel data merged with administrative records from the National Student Clearinghouse and the Educational Longitudinal Study (ELS) of 2002. Civilians were nearly three times as likely to enroll in college within one year of high school graduation (or one year of separation). Civilians were also much more likely to earn a bachelor’s degree within the period of study than either of the Army samples. While members of minority race/ethnicity groups in both military samples enroll at higher rates than their white counterparts, racial/ethnic minorities do not graduate at higher rates than their white counterparts. We discuss policy implications of our analyses in the final section of our paper.