The Scholastic Aptitude Test (SAT) is important for college admissions but its effectiveness and relevance are increasingly being questioned. This paper advances Integrated Control methods by introducing Transformed Control, a novel method that uses Large Language Models (LLMs) powered by Artificial Intelligence to construct control groups. We use OpenAI’s API to generate a control group where GPT-4, or ChatGPT, takes multiple SATs every year from 2008 to 2023. This control group helps analyze shifts in SAT math difficulty over time, beginning in the first year of 2008. Using the same trends. , we calculate the Average Difference in Scores (ADS) to examine changes in high school students’ math performance. Our results show a significant decrease in the difficulty of the math section of the SAT over time, which corresponds to a decrease in students’ math performance. The analysis shows a 71-point drop in SAT math difficulty from 2008 to 2023, and a 36-point drop in student performance, resulting in a total difference of 107 points in student math performance. We investigate possible mechanisms for this decline in math literacy, such as changing university preferences, increased screen time, inflation, and poor youth mental health. The difference between the math groups shows a decrease of 104 points for White students, 84 points for Black students, and 53 points for Asian students. Male students dropped 117 points, while female students dropped 100 points. This study highlights the need to rethink the role of the SAT in admissions and revise academic strategies to improve high school math performance.
That appears in a new paper by Saannidhya Rawat and Vikram K. Suresh, with Kevin Lewis leading the way.
Source link