Skip to main content

Table 1 Simulation Study Comparing BNSV, CLM, Frailty and GLMM Models: Posterior Mean Marginal Slope Estimates, Bias (Relative Bias), Variance and Mean Squared Error (MSE)

From: A Bayesian natural cubic B-spline varying coefficient method for non-ignorable dropout

  Normal (i) Normal (ii)
Method Slope Bias Var. MSE Slope Bias Var. MSE
BNSV -0.64 0.06(9%) 0.01 0.02 0.35 -0.002 (-1%) 0.01 0.01
CLM -0.45 0.25 (36%) 0.003 0.06 0.30 -0.05 (-14%) 0.03 0.03
Frailty -0.52 0.19 (27%) 0.004 0.04 0.16 -0.19 (-54%) 0.01 0.04
GLMM -0.21 0.49 (70%) 0.001 0.24 0.65 0.31(89%) 0.002 0.10
  Binary (i) Binary (ii)
Method Slope Bias Var. MSE Slope Bias Var. MSE
BNSV 7.49 0.13 (2%) 0.19 0.20 6.67 -0.05 (-1%) 0.17 0.18
CLM 8.02 0.66 (9%) 0.11 0.54 6.47 -0.24 (-4%) 0.14 0.20
Frailty 8.14 0.78 (11%) 0.14 0.75 6.04 -0.96 (-14%) 0.18 1.11
GLMM 9.09 1.74 (24%) 0.09 3.11 8.56 1.85 (28%) 0.14 3.55