TABLE 2.

Credibility interval coverage and convergence rates

	% in posterior 95% C.I.						% convergence failure
Parameter	1	2	3	4	5	6	1	2	3	4	5	6
	h2 = 0.56
σrep	0.0	81.8	0.0	18.8	23.8	60.0	0.0	0.0	0.4	0.0	0.8	0.8
σbl	96.0	94.8	97.0	95.0	6.4	12.0	0.0	0.0	36.6	6.2	0.0	0.4
σSCA	86.6	87.4	33.0	81.8	90.4	88.0	0.2	0.0	13.0	3.2	0.0	0.4
σgca	70.8	71.6	47.6	67.8	63.4	62.4	0.0	0.0	3.2	0.2	0.0	0.4
h2	96.2	85.6	11.2	78.6	93.8	89.8	0.0	0.0	2.8	0.2	0.0	0.2
βinbr	93.0	93.2	96.2	93.8	93.2	94.2	0.0	0.0	24.6	4.0	0.0	0.0
βoutc	96.4	97.0	97.2	96.6	95.4	94.8	0.4	0.0	24.2	5.8	0.0	0.2
μline	88.6	88.7	89.2	88.5	89.6	89.1	0.0	0.0	14.2	2.0	0.0	0.2
	h2 = 0.07
σrep	72.0	95.6	2.0	28.6	0.2	28.6	0.4	0.4	1.0	0.0	0.0	0.8
σbl	95.8	95.6	96.6	94.6	6.2	18.4	1.8	2.6	24.8	11.0	0.0	0.0
σSCA	28.0	36.0	0.4	12.0	89.6	79.6	0.6	1.8	9.4	3.2	0.2	0.0
σgca	0.8	1.0	0.4	0.4	0.0	0.0	0.6	0.0	4.2	1.2	0.2	0.0
h2	5.8	2.2	65.2	22.2	1.4	2.4	0.6	0.2	3.6	0.6	0.2	0.0
βinbr	96.4	96.2	98.0	96.6	95.6	95.2	0.6	1.0	12.4	3.4	0.0	0.0
βoutc	98.8	98.8	98.4	97.2	96.0	95.6	0.8	2.0	12.0	3.4	0.2	0.0
μline	91.5	91.3	92.9	92.5	92.2	91.8	0.5	0.6	6.7	1.2	0.2	0.0

Models 1–6 are as in Figure 2. The set of columns on the left gives fractions of all simulations where the true value of a given parameter lies in the posterior 95% credibility interval of its estimate. The right set of columns reflects how often the Gelman–Rubin diagnostic (see methods) exceeds 1.5. Other threshold values lead to the same conclusions. Data for line means of all lines were pooled.