Table 4.
N = 250 | N = 500 | N = 1000 | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Type | K | Method | Bias | SD | rMSE | Com | Bias | SD | rMSE | Com | Bias | SD | rMSE | Com |
– | 1 | CD | 0.004 | 0.179 | 0.179 | 0 | 0.004 | 0.127 | 0.127 | 0 | 0.004 | 0.089 | 0.089 | 0 |
CC | 0.363 | 0.240 | 0.435 | 0 | 0.365 | 0.167 | 0.401 | 0 | 0.365 | 0.117 | 0.383 | 0 | ||
U | 5 | iMICE | 0.067 | 0.258 | 0.267 | 0 | 0.037 | 0.176 | 0.180 | 0 | 0.022 | 0.121 | 0.123 | 0 |
avgmMICE | 0.011 | 0.218 | 0.218 | 1290 | 0.006 | 0.151 | 0.151 | 1290 | 0.004 | 0.106 | 0.106 | 1290 | ||
cslMICE | 0.005 | 0.214 | 0.214 | 1935 | 0.004 | 0.151 | 0.151 | 1935 | 0.004 | 0.106 | 0.106 | 1935 | ||
siMICE | 0.004 | 0.213 | 0.213 | 1290 | 0.004 | 0.150 | 0.150 | 1290 | 0.004 | 0.106 | 0.106 | 1290 | ||
10 | iMICE | 0.146 | 0.286 | 0.321 | 0 | 0.080 | 0.201 | 0.216 | 0 | 0.047 | 0.138 | 0.146 | 0 | |
avgmMICE | 0.022 | 0.221 | 0.222 | 1290 | 0.011 | 0.153 | 0.154 | 1290 | 0.006 | 0.106 | 0.106 | 1290 | ||
cslMICE | 0.005 | 0.215 | 0.215 | 1935 | 0.004 | 0.150 | 0.150 | 1935 | 0.005 | 0.106 | 0.106 | 1935 | ||
siMICE | 0.004 | 0.213 | 0.213 | 1290 | 0.004 | 0.150 | 0.150 | 1290 | 0.005 | 0.105 | 0.105 | 1290 | ||
E | 5 | iMICE | 0.027 | 0.217 | 0.219 | 0 | 0.014 | 0.151 | 0.152 | 0 | 0.009 | 0.105 | 0.106 | 0 |
avgmMICE | 0.013 | 0.215 | 0.215 | 1290 | 0.006 | 0.151 | 0.151 | 1290 | 0.004 | 0.105 | 0.105 | 1290 | ||
cslMICE | 0.019 | 0.243 | 0.244 | 1935 | 0.006 | 0.156 | 0.156 | 1935 | 0.005 | 0.107 | 0.108 | 1935 | ||
siMICE | 0.005 | 0.213 | 0.213 | 1290 | 0.005 | 0.150 | 0.150 | 1290 | 0.004 | 0.105 | 0.105 | 1290 | ||
10 | iMICE | 0.076 | 0.228 | 0.240 | 0 | 0.032 | 0.152 | 0.155 | 0 | 0.016 | 0.106 | 0.108 | 0 | |
avgmMICE | 0.024 | 0.219 | 0.221 | 1290 | 0.012 | 0.151 | 0.152 | 1290 | 0.006 | 0.106 | 0.106 | 1290 | ||
cslMICE | – | – | – | – | 0.020 | 0.194 | 0.195 | 1935 | 0.006 | 0.114 | 0.114 | 1935 | ||
siMICE | 0.005 | 0.213 | 0.213 | 1290 | 0.004 | 0.150 | 0.150 | 1290 | 0.004 | 0.105 | 0.105 | 1290 |
Reported are Bias, average bias; SD, Monte Carlo standard deviation; rMSE, root mean squared error; Com, number of communications. The cslMICE method failed in a few cases due to instability when N = 250 samples are evenly (E) distributed over K = 10 sites. N is the total number of samples. See Table 1 for local sample sizes. Results are based on 1000 Monte Carlo datasets.