Table 2.
Performance of methods for regression of (continuous) distress at Wave II, with missing baseline distress
|
% complete data |
Method |
Diet (β1 = −0.101) |
Distress (β2 = 0.554) |
Health (β6 = 0.042) |
|||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Bias | SE | StdBias | Coverage | Bias | SE | StdBias | Coverage | Bias | SE | StdBias | Coverage | ||
| 90% |
CCA |
−0.016 |
0.069 |
−0.229 |
93.9% |
<0.001 |
0.036 |
0.014 |
95.1% |
0.002 |
0.032 |
0.063 |
93.7% |
| |
MVNI-log |
−0.001 |
0.064 |
−0.012 |
93.0% |
−0.009 |
0.035 |
−0.252 |
95.1% |
0.003 |
0.031 |
0.095 |
94.1% |
| |
MVNI-skew0 |
−0.001 |
0.064 |
−0.010 |
93.2% |
<0.001 |
0.035 |
0.001 |
95.4% |
0.002 |
0.031 |
0.061 |
94.2% |
| 75% |
CCA |
−0.031 |
0.082 |
−0.383 |
92.4% |
−0.009 |
0.040 |
−0.219 |
95.4% |
0.001 |
0.036 |
0.016 |
95.3% |
| |
MVNI-log |
0.004 |
0.067 |
0.059 |
95.1% |
−0.032 |
0.037 |
−0.863 |
84.7% |
0.003 |
0.032 |
0.110 |
94.7% |
| |
MVNI-skew0 |
0.003 |
0.066 |
0.047 |
94.5% |
−0.009 |
0.038 |
−0.233 |
96.0% |
0.001 |
0.031 |
0.033 |
94.5% |
| 50% |
CCA |
−0.054 |
0.118 |
−0.458 |
91.9% |
−0.011 |
0.051 |
−0.210 |
95.5% |
0.003 |
0.048 |
0.064 |
94.9% |
| |
MVNI-log |
0.002 |
0.075 |
0.031 |
94.6% |
−0.064 |
0.045 |
−1.426 |
68.7% |
0.008 |
0.034 |
0.246 |
94.5% |
| |
MVNI-skew0 |
<0.001 |
0.074 |
−0.003 |
95.2% |
−0.012 |
0.046 |
−0.263 |
94.7% |
0.002 |
0.033 |
0.050 |
94.8% |
| 25% |
CCA |
−0.058 |
0.190 |
−0.308 |
91.8% |
−0.020 |
0.075 |
−0.265 |
93.7% |
0.002 |
0.071 |
0.031 |
94.8% |
| |
MVNI-log |
0.004 |
0.091 |
0.041 |
95.4% |
−0.106 |
0.060 |
−1.775 |
57.4% |
0.013 |
0.039 |
0.328 |
93.9% |
| |
MVNI-skew0 |
0.004 |
0.092 |
0.045 |
94.6% |
−0.023 |
0.059 |
−0.397 |
90.7% |
0.003 |
0.039 |
0.072 |
94.4% |
| 10% |
CCA |
−0.023 |
0.345 |
−0.068 |
94.8% |
−0.025 |
0.125 |
−0.196 |
95.9% |
0.008 |
0.120 |
0.071 |
96.0% |
| |
MVNI-log |
0.011 |
0.129 |
0.084 |
96.0% |
−0.145 |
0.092 |
−1.576 |
60.9% |
0.021 |
0.049 |
0.423 |
91.3% |
| MVNI-skew0 | 0.013 | 0.136 | 0.096 | 95.1% | −0.056 | 0.091 | −0.614 | 89.5% | 0.009 | 0.052 | 0.183 | 94.8% | |
Measures of performance are mean values from the estimation of the β parameters in Equation 1 across the 1000 simulated datasets of 1000 observations (compared to the true values from the synthetic population of 971,327). CCA = Complete Case Analysis; MVNI = Multivariate normal imputation; StdBias = standardised bias; SE = average (estimated) standard error across the 1000 datasets.