Table 2. Model metrics, impact of non-pharmaceutical interventions, COVID-19 pandemic, Europe, 2020.

Model	Deviance information criterion	Watanabe–Akaike information criterion	Conditional predictive ordinate	Dispersion
Cases	18,009.4	18,012.6	−9,006.6	1.01
Deaths	8,032.4	8,035.9	−4,018.4	0.89

COVID-19: coronavirus disease.

The Watanabe–Akaike information criterion (W-AIC) is described by Watanabe in 2010 [46] and was developed to specifically help identify best model fit in Bayesian models. Smaller W-AIC values mean better fit compared with alternative model specifications. The conditional predictive ordinate is a Bayesian diagnostic that detects surprising observations [47].