Skip to main content
. Author manuscript; available in PMC: 2022 Sep 30.
Published in final edited form as: Stat Med. 2021 May 14;40(22):4751–4763. doi: 10.1002/sim.8999

TABLE 4.

The bias and coverage of sensitivity and specificity for each of the four tests across the various simulation strategies

M = 2
 M = 3
Bias
 Coverage
 Bias
 Coverage
N = 225 N = 1000 N = 225 N = 1000 N = 225 N = 1000 N = 225 N = 1000
sens1 0.005 −0.004 0.92 0.90 −0.613 −0.617 0.00 0.00
sens2 0.007 −0.002 0.92 0.92 0.235 0.227 0.06 0.00
sens3 −0.092 −0.041 0.94 0.92 0.623 0.677 0.00 0.00
sens4 0.011 0.006 0.98 0.91 0.628 0.629 0.00 0.00
spec1 −0.001 −0.000 0.89 0.90 −0.042 −0.040 0.31 0.01
spec2 0.000 −0.000 0.90 0.89 0.034 0.033 0.68 0.17
spec3 −0.007 0.000 0.97 0.92 0.118 0.123 0.03 0.00
spec4 0.005 0.005 0.95 0.91 0.106 0.103 0.00 0.00
π 1 0.012 0.009 0.97 0.92 0.082 0.076 0.26 0.01

Note: The value of M denotes the number of states in the true model. The labels sensi and speci represent the sensitivity and specificity of test i, respectively. The posterior median was used for estimation and the coverage is based off a 90% central credible interval.