There are numerous reports that colleges use proprietary mathematical algorithms as part of their process for evaluating applications. Some colleges hire statistical experts known as "enrollment consultants" to help them predict enrollment by developing computer models to select applicants in such a way as to maximize yield and acceptance rates. Some of these models take into account factors such as an applicant's "zip code, religion, first-choice major and extracurricular interests, as well as academic performance". Colleges have been reported to have mathematical algorithms that recalculate an applicant's high school grade point average by weighing different course grades by factors such as perceived course difficulty and strength of high school curriculum This helps the college come up with a revised GPA number for a student to compare against applicants from other schools. Furthermore, many colleges track how well other students from the same high school have done--that is, applicants from the same high school who attended or are attending the college--by comparing their high school grades against their college grades, and admissions officers use this data to try to estimate the likely college grade performance of a given applicant. Generally admissions departments do not reveal the particulars of such mathematical analyses. According to Michele Hernandez, Ivy League admissions departments compile an academic index based on three main factors:
Highest SAT Reasoning Math/Critical Reading score
Average of three highest SAT Subject tests
Converted rank score based on grades, class rank, and high school difficulty
In her view, two-thirds of this evaluation is based on tests, while only one third is based on grades, leading her to conclude at one point that grades were less important overall as a factor than test scores, while in a different chapter she also suggested that the high school transcript information described roughly 60% of the college's perception of a student's academic performance. Next, the composite academic index score was combined with an analysis of personal factors such as extracurricular activity or the essay, such that the academic factor was weighted 70% to 85% while the personal evaluation was weighted only 30% to 15%. Generally the particulars of the mathematical formulas are not revealed to the public, and different colleges have different formulas. Part of the purpose of algorithms is to expedite the handling of thousands of applications in a short amount of time. For example, at Dartmouth College, data goes into a master card for each application, which leads to a ready sheet, where readers summarize applications; then, an initial screening is done: top applications go directly to the director of admissions for approval while lackluster ones go to another director. Dartmouth uses "A" for accept, "R" for reject, "P" for possible, with "P+" and "P-" being variants. A committee might spend a week with the "P" ones, of which they only accept about a sixth, according to Hernandez.