Schools at every level of selectivity are using predictive models. While the inputs can and do vary, many organizations (e.g., Collegeboard, ACT, enrollment strategy consultants) gather data, sell reports and/or take on private projects designed to inform these models.
On another thread there was much discussion about Collegeboard’s Environmental Context report. This report is sold to colleges to help them assess the likely financial picture of applicants. https://secure-media.collegeboard.org/digitalServices/pdf/professionals/data-driven-models-to-understand-environmental-context.pdf
Just as data mining has impacted many other industries, it is no surprise that it is impacting college admissions and enrollment. The predictive models of some schools include applicant financial inputs whether the school is need blind or need aware. No matter the schools endowment, every school has an annual financial aid budget that they try and stick to, they can’t just reach into the endowment when they want to spend more on fin aid in a given year. So, enter predictive models which help the school understand which applicants to give offers of admissions to, which students are most likely to enroll, and how much each applicant is likely to be able to pay, among other things.
Going over total operational budget for a number of years in a row is exactly what Middlebury is trying to address by their ongoing financial sustainability effort…which includes cost cutting, forced professor retirements and increasing student enrollment, to highlight a few initiatives.