Structural equation model for the causal relationship between the variables at the medical school level. Path strengths are shown as ∃ (standardized) path coefficients and significance levels based on a t-statistic with 17 degrees of freedom. The width of paths is proportional to the path coefficient. The saturated model allowed all variables to the left of a variable to have a causal influence on that variable and non-significant paths were removed until paths remaining were significant with p < 0.05. Paths not shown as causal arrows did not reach significance with p < 0.05.
McManus et al. BMC Medicine 2008 6:5 doi:10.1186/1741-7015-6-5