AI and Holistic Review

What is it about?

Selective colleges and universities use a system called "holistic review" in their admissions protocols, meaning that instead of just considering GPA or test scores, they consider other factors about the applicant such as their extracurricular activities, letters of recommendation, and personal statements. However, the Harvard admissions case showed that there is a potential for bias in these materials as Asian/Asian-American applicants received lower scores about their personality than other applicants. We propose the use of AI to analyze personal statements submitted by applicants as a way to help train admissions personnel to read with less bias. This study found that admissions essays are highly gendered (~80% classification accuracy for predicting male/female gender from essays) and classed (~70% classification accuracy for predicting above/below median income from essays). These results could help prime the reading of essays and other application materials.

Featured Image

Photo by Jess Bailey on Unsplash

Photo by Jess Bailey on Unsplash

Why is it important?

Many colleges and universities require essays as part of their applications, but little research has been done on how they are read, potential for bias in reading, and demographic patterning in the essays. We find that the essays are highly predictive of applicant gender and social class and propose our approach as a way to show admissions stakeholders what to be careful of in their reading of the essays. The covid-19 pandemic has only increased the importance of this work as many schools have gone test optional but have not gone essay optional.