Skip to main content

Table 3 Relative bias percentage, RMSE and coverage probability for methamphetamine use comparing validation data to imputed data

From: Comparing single and multiple imputation strategies for harmonizing substance use data across HIV-related cohort studies

Method

% Missing

Estimate

Mean Bias

% Relative Bias

RMSE

Coverage

Missing Data Mechanism: MCAR

 LD

10%

37.4%

-0.00055

-0.15%

0.0105

95.6%

 LR

10%

37.5%

-0.00031

-0.08%

0.0105

94.2%

 HD

10%

37.4%

-0.00035

-0.09%

0.0108

94.0%

 MI (M = 5)

10%

37.5%

-0.00013

-0.03%

0.0115

95.0%

 MI (M = 20)

10%

37.5%

-0.00013

-0.04%

0.0128

95.2%

 LD

30%

37.5%

-0.00003

-0.01%

0.0117

95.2%

 LR

30%

37.4%

-0.00035

-0.09%

0.0110

93.2%

 HD

30%

37.5%

-0.00017

-0.04%

0.0115

92.2%

 MI (M = 5)

30%

37.5%

0.00040

0.11%

0.0115

95.4%

 MI (M = 20)

30%

37.5%

0.00021

0.06%

0.0128

95.4%

 LD

50%

37.4%

-0.00064

-0.17%

0.0148

94.6%

 LR

50%

37.5%

-0.00005

-0.01%

0.0117

89.4%

 HD

50%

37.5%

0.00002

0.01%

0.0128

87.8%

 MI (M = 5)

50%

37.6%

0.00071

0.19%

0.0115

95.2%

 MI (M = 20)

50%

37.6%

0.00071

0.19%

0.0128

94.2%

Missing Data Mechanism: MAR

 LD

10%

34.9%

-0.02540

-6.78%

0.0276

30.8%

 LR

10%

36.7%

-0.00818

-2.18%

0.0133

83.8%

 HD

10%

37.3%

-0.00216

-0.58%

0.0108

93.0%

 MI (M = 5)

10%

37.5%

0.00034

0.09%

0.0104

94.6%

 MI (M = 20)

10%

37.5%

0.00036

0.10%

0.0103

94.4%

 LD

30%

30.8%

-0.06707

-17.89%

0.0680

0.0%

 LR

30%

35.7%

-0.01803

-4.81%

0.0210

56.0%

 HD

30%

37.1%

-0.00338

-0.90%

0.0121

88.6%

 MI (M = 5)

30%

37.5%

0.00044

0.12%

0.0110

95.6%

 MI (M = 20)

30%

37.5%

0.00054

0.14%

0.0109

96.0%

 LD

50%

26.9%

-0.10551

-28.15%

0.1063

0.0%

 LR

50%

35.2%

-0.02289

-6.11%

0.0257

37.4%

 HD

50%

37.1%

-0.00364

-0.97%

0.0134

86.6%

 MI (M = 5)

50%

37.6%

0.00082

0.22%

0.0120

94.4%

 MI (M = 20)

50%

37.6%

0.00076

0.20%

0.0118

93.0%

Missing Data Mechanism: MNAR

 LD

10%

35.0%

-0.02520

-6.72%

0.0274

32.6%

 LR

10%

35.5%

-0.01959

-5.22%

0.0223

50.2%

 HD

10%

35.8%

-0.01650

-4.40%

0.0197

61.6%

 MI (M = 5)

10%

35.8%

-0.01651

-4.41%

0.0196

61.4%

 MI (M = 20)

10%

35.8%

-0.01651

-4.40%

0.0196

62.2%

 LD

30%

35.0%

-0.02520

-6.72%

0.0274

32.6%

 LR

30%

35.5%

-0.01959

-5.22%

0.0223

50.2%

 HD

30%

33.5%

-0.04014

-10.71%

0.0417

3.0%

 MI (M = 5)

30%

35.8%

-0.01651

-4.41%

0.0196

61.4%

 MI (M = 20)

30%

33.5%

-0.03958

-10.56%

0.0410

3.6%

 LD

50%

27.9%

-0.09619

-25.66%

0.0971

0.0%

 LR

50%

31.6%

-0.05870

-15.66%

0.0598

0.0%

 HD

50%

32.4%

-0.05061

-13.50%

0.0521

0.8%

 MI (M = 5)

50%

32.6%

-0.04917

-13.12%

0.0505

2.2%

 MI (M = 20)

50%

32.6%

-0.04914

-13.11%

0.0504

1.6%

  1. Abbreviations, LD Listwise deletion, LR Logistic regression, HD Hot-deck, MI Multiple imputation, MCAR Missing completely at random, MAR Missing at random, MNAR Missing not at random, RMSE Root mean square error