model, and l0 is the log-likelihood of the null model with only the intercept as a predictor. The calculated R2 is consistent with the classical R2 and can be interpreted as the proportion of explained variation (Betts et al., 2006; Nagelkerke, 1991).