Analyzing Application Essays

Does language use and topic selection predict academic success in college? A new study suggests it does. (Hint to high schoolers: Move beyond personal experiences.)

January 8, 2015

Admissions essays are thought of by many as less scientific than other parts of the college application process -- a chance to share a personal story, to inject personality into the process, to become more than just a grade-point average or test score.

But it may be that statistical analysis can be applied to application essays -- and that some words and some topics correlate with better performance in college. That's the conclusion of a new study published in PLOS One that analyzes the words used in application essays with grades earned once enrolled.

The study found that the essays that predicted the most academic success demonstrated "categorical thinking," which involves writing that categorizes things, and that connects concepts and ideas. Generally, writing with categorical thinking uses many articles such as "the" and prepositions such as "on" and "of." 

Essays that show "dynamic thinking," in contrast, predict lower G.P.A.s in college. This writing tends to use pronouns such as "I" and "they" and to rely on personal narratives.

The authors of the paper -- all at the University of Texas at Austin -- are James Pennebaker, a psychology professor, David Beaver, professor in of linguistics; Gary Lavergne, program manager in the Office of Admissions; Cindy Chung, psychology postdoctoral fellow; and Joey Frazee, a linguistics graduate student.

The analysis is based on data from 50,000 essays from 25,975 applicants who, after being accepted, enrolled at "a large state university" from 2004 through 2007, and were then tracked for their grades. The study does not explicitly state that the students are at UT Austin, and the researchers declined to name the institution. But the size of the university seems to match UT, and the Institutional Review Board that reviewed the project was at that university.

Via email, Professor Beaver answered some questions about the study. While much of the analysis focused on relatively short words, the scholars also noticed trends with longer words, he said.

Generally, those applicants who, compared to the average applicant, used greater numbers of long words (6 letters or more) than others, used more complicated sentences, and wrote longer essays all ended up with slightly higher GPAs than did other admitted students.

Asked if applicants should thus avoid personal narratives, Beaver said it was too early to act on the new research because, to his knowledge, no admissions offices are analyzing applicants' essays based on this type of review.

"But I would advise students at the high school level that most colleges are more interested in writing about the world of ideas and things than in personal narratives," Beaver said. "Even when a college sets an essay title that seems to be asking for the applicants’ personal story, it’s a safe bet that the admissions officers (whether consciously or not) are looking for people who can put their own history in a broader context."

He said that he thinks college admissions officers should consider applying this study as one way to help predict applicant success. "The way people use language is one of many predictive factors a college admissions officer can use," he said. "Of course, admissions officers need to take many other factors into consideration when building a new incoming class, including the student’s specific interests and abilities, and all sorts of evidence of independence, maturity, and excellence in academics and beyond. But having said that, if you forced me at gunpoint to decide who to admit to the college, and gave me nothing to go on except the language measures developed in the new study, I’d do O.K. Based on the data I’ve looked at, and just considering the later academic performance of the students I picked, I wouldn’t do much worse than a professional admissions officer who had access to every student’s complete file."


Back to Top