Table 2.
Estimation results of mediation effects, expressed as Mean (MSE)
| Settings | MT methodsa | CONF methodsb* |
M1 = 0.16 (MSE) |
M2 = 0.20 (MSE) |
M3 = 0.25 (MSE) |
M4 = 0.30 (MSE) |
|---|---|---|---|---|---|---|
|
n = 300, p = 1000 |
HIMA | RA | 0.1547 (0.0043) | 0.1909 (0.0067) | 0.2414 (0.0065) | 0.2907 (0.0084) |
| PSR | 0.1628 (0.0048) | 0.1987 (0.0072) | 0.2520 (0.0073) | 0.3014 (0.0088) | ||
| IPW | 0.1581 (0.0054) | 0.1967 (0.0072) | 0.2451 (0.0071) | 0.2976 (0.0094) | ||
| OW | 0.1573 (0.0044) | 0.1947 (0.0069) | 0.2454 (0.0065) | 0.2955 (0.0083) | ||
| mHIMA2 | RA | 0.1488 (0.0041) | 0.1813 (0.0062) | 0.2320 (0.0062) | 0.2764 (0.0081) | |
| PSR | 0.1487 (0.0042) | 0.1850 (0.0064) | 0.2337 (0.0063) | 0.2837 (0.0083) | ||
| IPW | 0.1524 (0.0052) | 0.1897 (0.0068) | 0.2367 (0.0069) | 0.2884 (0.0094) | ||
| OW | 0.1522 (0.0043) | 0.1879 (0.0064) | 0.2369 (0.0064) | 0.2871 (0.0083) | ||
|
n = 500, p = 1000 |
HIMA | RA | 0.1596 (0.0024) | 0.2016 (0.0038) | 0.2458 (0.0040) | 0.3040 (0.0046) |
| PSR | 0.1684 (0.0027) | 0.2110 (0.0044) | 0.2563 (0.0044) | 0.3158 (0.0053) | ||
| IPW | 0.1607 (0.0026) | 0.2024 (0.0041) | 0.2470 (0.0045) | 0.3051 (0.0052) | ||
| OW | 0.1594 (0.0024) | 0.2019 (0.0038) | 0.2459 (0.0041) | 0.3041 (0.0046) | ||
| mHIMA2 | RA | 0.1540 (0.0022) | 0.1936 (0.0037) | 0.2349 (0.0040) | 0.2902 (0.0043) | |
| PSR | 0.1548 (0.0023) | 0.1966 (0.0037) | 0.2392 (0.0040) | 0.2977 (0.0045) | ||
| IPW | 0.1569 (0.0025) | 0.1980 (0.0040) | 0.2413 (0.0044) | 0.2994 (0.0050) | ||
| OW | 0.1556 (0.0023) | 0.1976 (0.0038) | 0.2402 (0.0040) | 0.2986 (0.0045) | ||
|
n = 300, p = 10,000 |
HIMA | RA | 0.1217 (0.0045) | 0.1514 (0.0061) | 0.1873 (0.0088) | 0.2278 (0.0119) |
| PSR | 0.1472 (0.0044) | 0.1823 (0.0050) | 0.2256 (0.0064) | 0.2750 (0.0087) | ||
| IPW | 0.1569 (0.0048) | 0.1935 (0.0058) | 0.2363 (0.0066) | 0.2887 (0.0086) | ||
| OW | 0.1567 (0.0043) | 0.1929 (0.0051) | 0.2358 (0.0061) | 0.2875 (0.0078) | ||
| mHIMA2 | RA | 0.1259 (0.0039) | 0.1540 (0.0053) | 0.1896 (0.0076) | 0.2307 (0.0102) | |
| PSR | 0.1350 (0.0039) | 0.1698 (0.0049) | 0.2089 (0.0064) | 0.2565 (0.0083) | ||
| IPW | 0.1425 (0.0043) | 0.1780 (0.0055) | 0.2169 (0.0066) | 0.2692 (0.0086) | ||
| OW | 0.1425 (0.0039) | 0.1780 (0.0049) | 0.2170 (0.0062) | 0.2670 (0.0080) | ||
|
n = 500, p = 10,000 |
HIMA | RA | 0.1502 (0.0022) | 0.1911 (0.0037) | 0.2364 (0.0040) | 0.2836 (0.0055) |
| PSR | 0.1596 (0.0024) | 0.2016 (0.0037) | 0.2497 (0.0038) | 0.2984 (0.0050) | ||
| IPW | 0.1577 (0.0023) | 0.2008 (0.0038) | 0.2502 (0.0037) | 0.2985 (0.0052) | ||
| OW | 0.1573 (0.0021) | 0.2008 (0.0034) | 0.2495 (0.0035) | 0.2991 (0.0047) | ||
| mHIMA2 | RA | 0.1287 (0.0024) | 0.1630 (0.0037) | 0.2022 (0.0046) | 0.2420 (0.0067) | |
| PSR | 0.1402 (0.0021) | 0.1805 (0.0032) | 0.2243 (0.0035) | 0.2719 (0.0048) | ||
| IPW | 0.1461 (0.0022) | 0.1872 (0.0036) | 0.2333 (0.0035) | 0.2800 (0.0052) | ||
| OW | 0.1456 (0.0020) | 0.1871 (0.0031) | 0.2324 (0.0033) | 0.2808 (0.0047) |
a MT methods denote two different mediation test approaches, including HIMA and modified HIMA2 (termed mHIMA2).
b CONF methods denote different confounding adjustment methods
* RA denotes regression adjustment. PSR denotes propensity score regression adjustment. IPW denotes inverse probability weighting. OW denotes overlapping weighting