Associate Vice President PREDICTIVE MODELING Daniel V. Pinch III Associate Vice President Emerson College
WHY? To reduce your discount rate In order to attract and retain higher quality students Attract higher quality students while keeping discount rate constant
Which category does your tuition fall into? I. “High Interest” Institution: Many quality applicants to choose from “Mild Interest” Institution: A good number of applicants to choose from “Low Interest” Institution: A select group of applicants to choose from
Do you know your institutions current acceptance and yield rates?
Many individuals within the institution do not/cannot visualize a problem Students are expecting more aid Federal limits on aid Institutional aid is capped as a % of revenues Institutional leaders and decision makers need to proactively spend money to identify and rectify problems
Enrollment is up Housing is full Retention is high In This Example . . . Enrollment is up Housing is full Retention is high
What is the Problem?
~Many solutions will be offered~ Once a Problem has been identified The etiology is often contested Decision makers have differing views on the causes of the problem -Diversity -Retention of key students -Increased competition Leading to different ideas on potential solutions ~Many solutions will be offered~
The challenge is to identify a workable solution ‘Vendor selection is the key to solving the problem. The right vendor is truly an invested partner.’
The following items must be worked out . . . Data collection Internal vs. External Data Ownership Compatibility with team Belief in the proposed solution Success measures Solution costs
Admission vs. Financial aid Deadlines Effective project scheduling Project point person Client input sessions Implementation Deliverables Review sessions
Plan on Problems Formatting of data Timing of transmissions Terminology
Aid allocation Process In order to better understand the aid allocation process as it impacts the enrollment continuum, the institution identifies 4 mutually exclusive groups: Accepted - Enrolled - Withdrawn - Graduated
Profiles were done on a variety of key student segments Gender State Residency SAT Scores Entry Date Academic Department High School Rank
We looked at the aid students only In this model
~ Once the students most likely to enroll and persist are identified the data is placed into a model ~
~ The model will put the students into 10 Decile’s; each identified by their likelihood of enrolling and persisting ~
The model was implemented and was VERY successful!