Analyzing Learning Assistant influence on STEM student success using logistic and hierarchical regression

The increasing demand for a robust science, technology, engineering, and mathematics (STEM) workforce highlights the need to understand factors that enhance student success in STEM fields. Despite significant need for STEM-qualified individuals, less than half of students initially expressing STEM interest upon college entry graduate with a STEM degree, dropping lower for underrepresented students. The Learning Assistant (LA) program, implemented at colleges around the world, involves students (LAs) aiding their peers through evidence based collaborative activities in STEM courses. It has been well documented that LAs are associated with short-term student success (lower course failure rates) and long-term student success (higher graduation rates). In this study we investigate the impact of the LA program on student success in introductory STEM courses. We analyzed over 10 years of student data, focusing on DFW (D, F, or withdraw) and six-year graduation rates. Using logistic regression and hierarchical linear models, we assessed the influence of LA support on student outcomes, with particular attention to marginalized demographics and repeated LA exposure. We show that LA-supported students in introductory physics courses experienced a 7% decrease in DFW rates. Notably, underrepresented students saw a 10% reduction in DFW rates. Additionally, repeated LA exposure in physics courses provided greater benefits for DFW rates compared to single-course exposure. This research underscores the importance of LA programs in improving STEM education outcomes, notably for underrepresented students.