Table 2.

Data‐based predictive distributions for heterogeneity variances ${\tau }_{\mathit{kl}}^2$, by outcome type, and corresponding distributions for the ${\tau }_k^2$

	Predictive Distribution for the ${\mathit{\tau }}_{\mathit{k}\mathit{l}}^{\mathbf{2}}$	Corresponding Distribution for the ${\mathit{\tau }}_{\mathit{k}}^{\mathbf{2}}$,
		Based on Matching Moments in (6), a
All‐cause mortality	LN(−4.28, 1.612)	LN(−4.83, 1.692)
Obstetric outcomes	LN(−3.33, 1.602)	LN(−3.88, 1.692)
Cause‐specific mortality/major morbidity	LN(−3.52, 1.612)	LN(−4.08, 1.702)
event/composite (mortality or morbidity)
Resource use/hospital stay/process	LN(−2.21, 1.602)	LN(−2.76, 1.692)
Surgical/device‐related success/failure	LN(−1.86, 1.612)	LN(−2.42, 1.702)
Withdrawals/dropouts	LN(−2.85, 1.602)	LN(−3.40, 1.692)
Internal/external structure‐related outcomes	LN(−2.53, 1.612)	LN(−3.09, 1.702)
General physical health indicators	LN(−2.37, 1.612)	LN(−2.93, 1.702)
Adverse events	LN(−1.97, 1.602)	LN(−2.52, 1.692)
Infection/onset of new disease	LN(−2.55, 1.602)	LN(−3.10, 1.692)
Signs/symptoms reflecting continuation/	LN(−2.13, 1.602)	LN(−2.68, 1.692)
end of condition
Pain	LN(−1.85, 1.602)	LN(−2.40, 1.692)
Quality of life/functioning (dichotomised)	LN(−2.59, 1.622)	LN(−3.15, 1.712)
Mental health indicators	LN(−2.20, 1.622)	LN(−2.76, 1.712)
Biological markers (dichotomised)	LN(−1.83, 1.602)	LN(−2.38, 1.692)
Subjective outcomes (various)	LN(−2.75, 1.612)	LN(−3.31, 1.702)

^a Assuming that priors for the correlations ρ_km have mean 0.5 and variance 0.07, as when using the priors listed in Table 3.