The Link to the study can be found here
From the Abstract:
“Results show that essays have a stronger correlation to reported household income than SAT scores.”
Seems like there is so much animosity directed at SATs for their correlation with wealth when essays are even worse. Wonder what the solution, if there is one, is?
That’s not surprising at all.
I agree this is not surprising. One possible solution would be to require a graded HS paper, which some schools already do. But, this is an imperfect solution…I don’t think there are any perfect solutions out there.
From Page 6 of the report: “The relationship between essay content and SAT score is strongest for middle-income students and weakest for high-income students.”
Yep - no amount of test prep is going to turn a high income kid with a 1200 SAT into 1400+ scorer. But a one-on-one essay guide , costing upwards of $150/hr, can help turn a mediocre essay into a very good one. I live in an upper-middle class community and the parents here are overjoyed that more colleges are going test-optional. Median HS SAT score for the local HS is low relative to the SES of the community.
What makes you think that colleges put a high priority on choosing the most able/intelligent students for their classes? They have many other priorities (not least, for many of them, picking students who will pay more). Test optional gives them more freedom and even more importantly makes it harder to win the sorts of lawsuits we’ve seen against Harvard, because it’s harder to compare essays than SAT scores.
So why do you think colleges even want to find a “solution”? The move to get rid of test scores was gathering steam even before the pandemic and it just provided the perfect excuse.
Not all parents. If your kid has very high SAT scores you may/may not be happy about test optional as there are now more applicants and there will be more emphasis on essays, recs etc than test scores.
The linked page includes this:
“We quantify essay content using correlated topic modeling (CTM) and the Linguistic Inquiry and Word Count (LIWC) software package.”
How long before college admissions offices use something like this to auto-grade college application essays, even if just to check consistency between different admission readers?
Note that the study did not evaluate the admission readers’ raters of the essay or which students had the best essays from an admission stand point. Instead they estimated the essay topic by using a computer program that counted frequency of key words within the essay, and found that choosing certain essay topics (as estimated by the key word count) was correlated with income and SAT score.
For example, wring essays about “seeking answers” had a high 0.57 correlation with SAT score. “Seeking answers” topic was estimated by a high frequency of using key words like “question, book, like, research, read, answer, ask” within the essay. Similarly writing essays about “tutoring groups” shows a strong negative correlation with SAT score. Kids with higher SAT scores were less likely to write essays about “tutoring groups.”
The correlations between certain estimated essay topics and SAT score seem quite substantial, while the correlations between estimated essay topics and income seem much lower. The essay topic that was most correlated with income seems to be writing about China, as estimated by high frequency of the key words “chines, studi, student, also, time, china, school.” I’m not sure how much some of these key words have to do with China, but I don’t doubt that writing essays about China would be correlated with income – both due to higher income among Chinese international students (UC sample group) and among students who travel to China. However, this does not mean that the kids who wrote essays about China or traveling had better essays from an admission stand point-- only that they were more likely to have higher income.
As mentioned above, the study estimates essay topics by frequency of key words, then reviews how well different essay topics are correlated with income and SAT score. Unfortunately they don’t list the results by topic in a numerical table format – only a graphical format in which colors correspond to magnitude, so I am estimating magnitude by color. A list of essay topics with the highest and lowest correlation with income and SAT score is below.
Estimated Essay Topics with Largest Positive Correlation to Income
Estimated Essay Topics with Largest Positive Correlation to Sat Score
Estimated Essay Topics with Largest Negative Correlation to Income
Estimated Essay Topics with Largest Negative Correlation to Sat Score
Not very long if the volume of applications continue to increase at this rate. If an average applicant doubles his/her number of applications, the overall number of applications also double. It’s a numbers game that humans just can’t keep up at some point.
It is my understanding (please correct me if not the case) that this does not have anything to do with admissions. Rather, it means that rich kids are more likely to have rich kid topics. The SAT has more of a direct effect on admissions.
As Data10 notes, this is related to essay content and topics, not rating by admissions counselors.
Why does “easily countable features of essays, like the number of commas they contain” mean “correlation with wealth” is “even worse” and "$150/hr, can help turn a mediocre essay into a very good one. "
From my read, the conclusion is that students in different SES groups write about different topics, using different words.