Table 2. Summary results for the external validation comparisons across 16,527 country-year data pairs for which a CBH and SBH estimate were both available.

We also show a subanalysis by data source, indicating good robustness of results across source type.

Age bin	SBH source	${\overline{\mathbf{q}}}_{\mathbf{a},\mathbf{c}\mathbf{b}\mathbf{h}}$	${\overline{\mathbf{q}}}_{\mathbf{a},\mathbf{s}\mathbf{b}\mathbf{h}}$	ME	SDE	MRE	MAPE	R2	R2corr
NN	All	0.030	0.033	0.0026	0.010	1.06	17.7%	0.52	0.96
Infant	All	0.061	0.064	0.0030	0.018	1.05	16.4%	0.64	0.96
Under-5	All	0.093	0.096	0.0029	0.028	1.04	16.4%	0.74	0.97
NN	MICS	0.038	0.041	0.0030	0.009	1.09	17.9%	0.5	0.96
Infant	MICS	0.075	0.080	0.0047	0.019	1.05	16.1%	0.67	0.97
Under-5	MICS	0.123	0.129	0.0055	0.029	1.05	16.3%	0.79	0.98
NN	Census	0.030	0.032	0.0019	0.010	1.02	18.4%	0.55	0.96
Infant	Census	0.062	0.063	0.0013	0.019	1.00	17.2%	0.68	0.97
Under-5	Census	0.097	0.096	−0.0005	0.026	0.98	17.0%	0.79	0.98
NN	Other	0.029	0.032	0.0029	0.010	1.07	17.5%	0.48	0.96
Infant	Other	0.056	0.059	0.0034	0.017	1.07	15.8%	0.57	0.95
Under-5	Other	0.083	0.087	0.0037	0.028	1.06	16.2%	0.62	0.96

Abbreviations: CBH, complete birth history; MAPE, median absolute percentage error; ME, mean error; MICS, Multiple Indicator Cluster Surveys; MRE, median relative error; NN, neonatal; $\overline{q}$, average estimated mortality probability; R², coefficient of determination; SBH, summary birth history; SDE, standard deviation of the errors.