Can Quality Advising Increase Bachelor’s Degree Attainment Among Low-Income Students?

The study asks whether intensive college advising helps low-income students earn bachelor’s degrees and why it works. The authors use data from a multisite randomized controlled trial, baseline surveys, advisor–student interaction records, and National Student Clearinghouse enrollment data. They find that advising increases four-year college enrollment by about 9 percentage points and raises bachelor’s degree completion by 7.6 percentage points within five years—about a 16 percent improvement over the control group. Most of this gain comes from students enrolling in higher-quality colleges, which explains roughly 70 percent of the increased degree attainment.

The Policy Scientist's Perspective

This article addresses a policy issue of broad significance: the persistent gap in college completion between low- and high-income students, a divide closely tied to long-term economic mobility. Its focus on advising is timely given growing concerns about declining social mobility and the limits of low-cost information interventions. The study’s randomized controlled trial, large multisite sample, and high-quality administrative data strengthen its credibility and support generalizability to similar urban settings. Its causal methods and clear evidence on college quality effects make it a meaningful contribution to the literature.

Full Citation and Link to Article

Barr, A., & Castleman, B. (2025). Increasing degree attainment among low-income students: The role of intensive advising and college quality. American Economic Review, 115(11), 4075–4103. https://doi.org/10.1257/aer.20240669

Central Research Question

The central question of this study is whether intensive college advising can meaningfully increase bachelor’s degree attainment among low-income students and, if so, through what mechanisms these gains arise. The authors are especially focused on whether advising works primarily by improving the quality of students’ initial college enrollment—specifically, whether it shifts students toward four-year institutions with higher graduation rates, stronger labor-market outcomes, and better overall institutional performance. A secondary but important part of the research question concerns the relative value of such advising compared with more limited informational “nudge” interventions, which have produced weak or inconsistent effects in prior work. The authors ask whether personalized, sustained advising represents a scalable and cost-effective strategy for reducing socioeconomic gaps in college completion.

Previous Literature

The study engages a well-developed body of research on college access, advising, and the role of institutional quality in shaping long-term educational and economic outcomes. Earlier studies establish that college quality matters substantially for both degree attainment and earnings, with selective or higher-performing institutions producing significantly stronger results even after accounting for student characteristics. At the same time, prior evaluations of advising programs have produced mixed evidence. Some studies have found modest increases in initial college enrollment, but many have identified fade-out in persistence or no meaningful effect on degree completion. Research on low-cost informational interventions—including text-message nudges, simplified FAFSA prompts, and basic outreach—generally shows limited or no long-term impact on enrollment or completion. The authors situate their study against this backdrop, noting the lack of rigorous evidence on whether more intensive and sustained advising can influence the full college trajectory, not just early decision points. This work also extends research on college-choice constraints, which shows that many high-achieving, low-income students under-apply to higher-quality institutions due to limited information, inaccurate beliefs about costs, or uncertainty about admission likelihoods. By incorporating baseline preference data and modern machine-learning methods to analyze treatment heterogeneity, the authors build upon the most cited empirical findings in this literature while addressing unresolved questions about mechanisms.

Data

The analysis relies on a uniquely rich combination of datasets. First, the authors use detailed application information from Bottom Line (BL), the advising program evaluated in the study. These data include demographic characteristics, academic indicators (e.g., GPA, test scores), family income, and baseline college preferences collected before randomization. The preference data are especially valuable, allowing the researchers to observe the set of institutions students considered before receiving any treatment. Second, the authors use a spring-of-senior-year survey that captures college applications, financial-aid behaviors, award-letter review, and students’ perceptions of the application process. They also analyze free-response survey answers to assess sentiment and the perceived difficulty of the college-choice process. Third, advisor–student interaction data provide detailed counts and timing of advising meetings across both the Access and Success components of the program. Finally, the authors use National Student Clearinghouse records to track college enrollment and degree attainment for up to six years after high school. The Clearinghouse data cover more than 97 percent of U.S. postsecondary institutions, which allows for near-complete tracking of educational trajectories. The dataset’s breadth, longitudinal structure, and inclusion of baseline preference information substantially strengthen the credibility and interpretability of the findings.

Methods

The study uses a multisite, multicohort randomized controlled trial involving more than 2,400 low-income students across several cities in Massachusetts and New York. Students meeting eligibility criteria—GPA of at least 2.5 and household income below 200 percent of the federal poverty line—were randomized either to receive an offer of intensive advising or to a control group that received no Bottom Line services. The analysis relies on intention-to-treat estimates, controlling for site-by-cohort fixed effects and a broad set of baseline covariates to increase precision. The authors estimate treatment effects on enrollment, enrollment quality, persistence, and bachelor’s degree attainment. A defining methodological innovation is the use of causal forest models to estimate personalized treatment effects for both initial enrollment quality and degree completion. These machine-learning methods provide more powerful and flexible tools for evaluating heterogeneous treatment effects than traditional subgroup analysis. The authors then incorporate the personalized estimates into an instrumental-variables mediation framework, allowing them to quantify the proportion of the advising effect on degree attainment that operates through improved college quality. They also conduct robustness checks, including alternative constructions of the baseline preference index, multiple model specifications, and analyses separating the pre-college and in-college advising components.

Findings/Size Effects

The results show large and statistically precise effects of intensive advising on both college enrollment patterns and bachelor’s degree attainment. Students offered advising were 9.1 percentage points more likely to enroll in a four-year college immediately after high school, a 13 percent increase over the control group. This shift was driven almost entirely by enrollment in higher-quality institutions, defined by graduation rates, loan-default performance, labor-market outcomes, and mobility metrics. Advising also reduced two-year college enrollment by about 4 percentage points. These initial quality improvements persisted over time, with treated students showing elevated four-year enrollment rates throughout the first four years after high school. Degree attainment effects are similarly substantial. Within five years of expected high-school graduation, bachelor’s degree attainment rose by 7.6 percentage points—roughly a 16 percent increase relative to the control mean. For the first cohort, the six-year effect was even larger at 9.6 percentage points. The authors find that roughly 70 percent of the treatment effect operates through improvements in initial college quality. The causal-forest analysis shows a strong positive correlation between individual-level treatment effects on enrollment quality and treatment effects on degree attainment, reinforcing the mechanism. The advising also appears to shift students toward using more reliable sources of support: treated students applied to nearly three additional colleges, were significantly more likely to review award letters with guidance, and expressed substantially lower frustration and confusion in survey responses.

Conclusion

The study provides strong causal evidence that intensive, personalized advising can raise bachelor’s degree attainment for low-income students at scale. Unlike previous advising interventions that generated modest or fading impacts, this program produces large and persistent improvements. The findings contribute meaningfully to the literature by showing that advising affects not just access but college choice, specifically pushing students toward higher-quality institutions with better completion rates. The careful mediation analysis offers rare empirical confirmation that college quality—not just enrollment—is the crucial mechanism. The combination of high-quality data, a large and diverse experimental sample, and rigorous methods increases confidence in both internal validity and potential generalizability to similar urban contexts. Overall, the study demonstrates that intensive advising can be one of the most cost-effective strategies available for increasing bachelor’s degree attainment among academically prepared, low-income students.