Skip to main content
. Author manuscript; available in PMC: 2020 Jul 17.
Published in final edited form as: Stat Methods Med Res. 2018 Jul 31;28(10-11):2897–2911. doi: 10.1177/0962280218790107

Table 1.

Performance of the proposed joint model compared with the joint frailty model and shared frailty models ignoring Level-2 clustering

N=300 Proposed joint model (PJM) Joint frailty model with Level-1 clustering only (JFM-L1) Shared frailty model with Level-1 clustering only (SFM-L1)
Parameter Estimate Bias SE SEM CP Estimate Bias SE SEM CP Estimate Bias SE SEM CP
β= −0.5 −0.483 3.4 0.215 0.190 0.944 −0.491 1.8 0.239 0.221 0.947 −0.327 52.9 0.153 0.152 0.787
α= −0.5 −0.502 0.4 0.262 0.284 0.958 −0.521 4.2 0.285 0.287 0.943 NA
γ2= 0.5 0.506 1.2 0.181 0.178 0.986 NA NA
γ1= 0.5 0.495 1.0 0.236 0.227 0.972 0.514 2.8 0.156 0.151 0.960 NA
θ2 =1.0 0.942 5.8 0.261 0.244 0.901 NA NA
θ1= 1.0 1.005 0.5 0.105 0.099 0.944 1.356 35.6 0.195 0.110 0.227 0.665 50.4 0.207 0.105 0.234
β= −0.5 −0.487 2.6 0.201 0.190 0.963 −0.501 0.2 0.221 0.221 0.950 −0.346 44.5 0.152 0.159 0.819
α= −0.5 −0.521 4.2 0.296 0.269 0.936 −0.473 5.4 0.297 0.288 0.940 NA
γ2= 1.0 1.060 6.0 0.242 0.207 0.938 NA NA
γ1= 0.5 0.535 7.0 0.191 0.205 0.951 0.719 43.8 0.172 0.157 0.680 NA
θ2 =1.0 0.929 7.1 0.219 0.239 0.901 NA NA
θ1= 1.0 0.989 1.1 0.089 0.099 0.988 1.348 34.8 0.187 0.113 0.223 0.673 48.6 0.230 0.117 0.305
β= −0.5 −0.505 1.0 0.209 0.200 0.952 −0.511 2.2 0.229 0.231 0.967 −0.307 62.9 0.167 0.163 0.768
α= −0.5 −0.480 4.0 0.282 0.251 0.912 −0.399 20.2 0.305 0.291 0.917 NA
γ2= 2.0 2.032 1.6 0.367 0.372 0.959 NA NA
γ1= 0.5 0.534 6.8 0.231 0.218 0.966 0.905 81.0 0.234 0.190 0.413 NA
θ2 =1.0 0.920 8.0 0.256 0.245 0.878 NA NA
θ1= 1.0 0.985 1.5 0.114 0.110 0.932 1.349 34.9 0.194 0.129 0.270 0.556 79.9 0.180 0.146 0.176

NA: Not Applicable; Percent Relative Bias = mean of |observedtruetrue|×100; SE = Sampling standard error of the parameter estimate; SEM = Sampling mean of the standard error estimate (SEM); CP = Coverage probability of the corresponding 95% confidence interval