Day:
Monday, November 4, 2019
Time:
1:30 PM - 2:30 PM
Location:
Emerald, Lobby Level
First Learning Outcome: Upon completion of this session, participants will be able to describe challenges of estimating effects of financial aid on enrollment.
Second Learning Outcome: Participants will learn about the visualization software that can be used to create the tool described during the session. A desktop version of this software can be downloaded and used for free.
Third Learning Outcome: Participants will learn about incorporating statistical models and "what if" scenarios in data visualizations.
Core Competencies: Interpretation and Application of Institutional and External Data, Professional Development and Contributions to the Field
Proficiencies: Admissions: Reporting Progress Toward Enrollment Goals, Enrollment Management: Developing Enrollment Mix
Intended Audience: Some experience in the profession, Significant experience in the profession
Visualizing Future Enrollment and Tuition Revenue
Category
Session
Description
Public universities increasingly use merit aid for enrollment management. By leveraging financial aid, institutions can attract students with desired academic and non-academic qualities. However, when used inefficiently, tuition discounting might lead to loss of revenue. Visualizing enrollment and tuition revenue for admitted applicants can help allocate financial aid awards and maximize tuition revenue. This presentation is focused on statistical models used to predict student enrollment decisions and on visualizing these predictions with a possibility to explore “what if” scenarios for financial aid offers.
The dataset for predictive models includes 71,462 students who were offered admission to a public university over the period of five years—fall 2013 through fall 2017. About 65% of admitted applicants are non-residents. International applicants are excluded. Because of the differences in tuition and in the probabilities of enrollment of in-state and out-of-state students, separate models are estimated for these groups. Including financial aid as a predictor of enrollment further complicates the analysis, since only some applicants—filers—submit FAFSA and are eligible for state and federally funded loans and grants. Furthermore, only some of the filers are eligible for Pell grants. Since each of these groups receives different financial aid package, separate models are estimated for non-filers, filers who are not eligible for Pell grants, and filers who are eligible for Pell grants.
Models for out-of-state students account for an applicant’s home state, because availability of options and scholarships for students in their home states vary by state. For in-state students, models account for a high school of an admitted applicant. The dependent variable for the models is the probability of enrollment, which is then used to predict tuition revenue. The expected tuition revenue is calculated by multiplying the probability of enrollment by the tuition and fees minus institutional financial aid. In addition to financial aid offers expressed as the percentage of tuition and fees, independent variables include gender, ethnicity, distance, college of interest, scholarship eligibility, first generation status, high school GPA and test scores, and legacy, such as an indicator of a sibling attending the institution or a relative who graduated from the institution in the past. Historical data are used to predict enrollment and one-year tuition revenue for 2018 cohort.
Most of the visualizations cover the entire applicant pool. An admission professional can then select an individual admitted applicant; a group of applicants by demographic characteristics, financial aid eligibility, residency, high school GPA and test scores, legacy, or geographic origin. Some visualizations provide an opportunity to review data and predictions by state of origin or by county and high school for in-state applicants.
Submission ID:
6419
Presenter(s):
Iryna Johnson Auburn University
Winner Status
- Session