This meta-analytic study 1) examines outcomes from primary research comparing the impact of problem-based learning (PBL) and traditional curricula on medical students' National Board of Medical Examiners (NBME) I and II performance and 2) explores the use of Hierarchical Linear Modeling (HLM) in identifying study and PBL implementation characteristics that predict these outcomes. NBME I and II overall scores were used as dependent variables, with study design (randomized/non-randomized), publication year, and PBL experience (number of years since the adoption of PBL as an instructional method at the time of the reported NBME results) as explanatory variables. Initial unconditional HLM (i.e., without inclusion of potential explanatory variables) results indicated PBL curricula had: 1) a positive, albeit not statistically significant, effect on NBME II, with an average effect size of 0.16; and 2) a negative effect on NBME I, with an average effect size of -0.15. Including explanatory variables in HLM analyses explained additional variability in NBME I effect sizes, and identified study design (gamma^1 = 0.82, p = 0.01) and PBL experience (gamma^3 = 0.07, p = 0.02) as significant predictors of positive PBL effects. Publication year, in contrast, had significant negative effects (gamma^2 = -0.06, p = 0.02). This study explained variability in NBME I effect sizes and clarified the impact of PBL discerned in previous reviews. Implications for future research include the need to examine curriculum features that operationally define PBL, as well as extend the consideration of outcomes on which PBL's impact can be examined.
Read full abstract