Association for Public Policy Analysis & Management

Pell receipt is widely used as a marker of low-income status in higher education research and reporting. However, Pell receipt has serious limitations as a proxy for low-income status, particularly for colleges that serve large numbers of low-income and under-represented students. This presentation examines these limitations within the context of research and reporting at the City University of New York (CUNY) and whether alternative measures better explain observed student-level variation in retention and completion.

The measure of Pell receipt is an imprecise proxy for low-income status. Instead of directly measuring individuals’ income levels, it marks those who have effectively accessed a support program (Romano & Millard, 2006). Students who are not citizens or permanent residents are not eligible, and the measure is dependent on a student’s ability to complete the application process. Furthermore, the population identified as low-income through Pell changes as eligibility criteria shift. Relatedly, Pell receipt increasingly encompasses a wide range of family income levels (Delisle, 2017; Cooper, 2018). These limitations of Pell receipt as a proxy for low-income status are particularly pronounced at CUNY, where a large majority (around 60 percent) of US resident undergraduates receive Pell grants. A larger percentage of the undergraduate population is likely poor enough to qualify for grants but fails to complete necessary paperwork or maintain Pell eligibility, and a substantial portion of additional non-recipients are non-citizens. Furthermore, as Pell eligibility has expanded, more middle-class students have joined poorer students as Pell recipients. As a result of these limitations, graduation rates between Pell recipients and non-recipients at CUNY are very similar and logistic regressions that predict graduation and control for other demographic and academic characteristics find that Pell-receipt is positively associated with persistence and graduation.

We explore alternative measures of low-income status and compare the results of analyses using Pell status and alternatives to predict college persistence and graduation. We find that the measure of Expected Family Contribution (EFC) from the FAFSA, which is the amount a family can be expected to contribute toward a student’s college costs, has several advantages over Pell status. Given that the majority of CUNY students receive Pell grants, the fact that the continuous measure of EFC allows more fine-grained categorization of low-income students is useful. Regression results find that lower EFC is associated with a lower probability of graduation (contrary to results using Pell status as a marker of low-income for the same population). However, even a dichotomous measure based on EFC, such as zero EFC and non-zero EFC, may be a more useful proxy for low-income because it identifies a needier population than Pell receipt (Davidson 2015). We examine whether the findings of Davidson 2015 on persistence hold at CUNY, as well as variation by EFC status for the longer-term outcome of graduation. We also use machine learning techniques to identify other meaningful divisions of EFC and reported family income.