Predicting medical student performance from attributes at entry: a latent class analysis

Area of Science:

• Medical Education Research
• Educational Psychology
• Biostatistics

Background:

• Student selection for medical degrees is critical for identifying candidates likely to succeed.
• Traditional selection methods may not fully capture the nuances of student performance trajectories.
• Understanding student typologies can optimize admissions processes and improve educational outcomes.

Purpose of the Study:

• To model typologies of medical student examination performance using prior academic achievement and interview ratings.
• To evaluate the utility of latent class analysis (LCA) in assessing the effectiveness of student selection.

Main Methods:

• Retrospective analysis of anonymised data from two cohorts of medical students (Bachelor of Medicine, Bachelor of Surgery).
• Application of latent class analysis (LCA) to identify student performance typologies.
• Triangulation of findings using logistic regression analysis.

Main Results:

• LCA identified three distinct student performance typologies based on prior academic achievement and interview scores.
• Prior academic achievement (especially in chemistry) and high interview scores were positively associated with successful examination performance.
• Logistic regression analysis supported the findings from LCA.

Conclusions:

• Latent class analysis (LCA) effectively provides meaningful data on the performance of selection processes.
• LCA serves as a valuable complementary tool to existing methods in educational research.
• The findings empirically inform the medical student selection process, potentially improving candidate identification and success rates.