Skip to main content

Table 2 Specification of the logistic regression models used to impose missing data under the two covariate-dependent MAR scenarios

From: The impact of missing data on analyses of a time-dependent exposure in a longitudinal cohort: a simulation study

  Odds ratio for missing (exp(γ i ))
γ Scenario 1 Scenario 2 $
1 (WC1, 10 cm) 1.10 1.21
2 (Age, years) 1.06 1.12
3 (Female) 1.10 1.21
4 (Education secondary) 0.72 0.52
5 (Education tertiary) 0.44 0.19
6 (COB UK) 1.15 1.32
7 (COB Mediterranean) 1.71 2.92
8 (Alcohol low) 0.77 0.59
9 (Alcohol moderate) 0.66 0.44
10 (Alcohol high) 0.85 0.72
11 (Smoking former) 1.16 1.35
12 (Smoking current) 1.80 3.24
13 (Physical low) 0.93 0.86
14 (Physical moderate) 0.99 0.98
15 (Physical high) 0.91 0.83
  1. $Odds ratio for Scenario2 = (Odds ratio for Scenario1) 2.