Skip to main content

Table 7 Comparison of approaches for the estimation of natural effects on the odds ratio scale for Scenario 2 with a binary mediator (based on 1000 data sets of size \(n=1000\))

From: An empirical evaluation of approximate and exact regression-based causal mediation approaches for a binary outcome and a continuous or a binary mediator for case-control study designs

Effect

Approach

True value

Mean

Bias

Relative bias (%)

SD

RMSE

CP (%) delta

CP (%) boot

NDE

Approx_Naive

3.529

5.629

2.100

59.510

1.197

2.417

41.3

35.1

Approx_C

3.529

4.890

1.361

38.566

1.011

1.696

68.7

64.0

Approx_IPW

3.529

5.039

1.510

42.795

1.061

1.846

63.5

56.5

Exact_Naive

3.529

2.733

-0.796

-22.544

0.399

0.890

55.0

57.4

Exact_IPW

3.529

3.617

0.088

2.504

0.545

0.552

95.1

94.9

Unified

3.529

4.897

1.368

38.767

1.011

1.701

71.5

-

NIE

Approx_Naive

1.460

2.486

1.026

70.293

0.254

1.057

0.0

0.0

Approx_C

1.460

1.097

-0.362

-24.816

0.151

0.392

57.3

35.5

Approx_IPW

1.460

1.581

0.121

8.281

0.151

0.194

97.2

87.6

Exact_Naive

1.460

1.929

0.469

32.154

0.135

0.488

1.5

1.2

Exact_IPW

1.460

1.472

0.013

0.869

0.116

0.117

99.0

95.6

Unified

1.460

1.097

-0.362

-24.823

0.150

0.392

51.4

-

TE

Approx_Naive

5.151

14.039

8.888

172.557

3.546

9.570

0.6

0.5

Approx_C

5.151

5.258

0.107

2.073

0.729

0.736

99.8

95.8

Approx_IPW

5.151

7.899

2.748

53.351

1.496

3.129

61.6

32.7

Exact_Naive

5.151

5.253

0.102

1.979

0.718

0.725

95.5

95.5

Exact_IPW

5.151

5.301

0.150

2.911

0.740

0.755

97.8

95.3

Unified

5.151

5.264

0.113

2.202

0.722

0.731

95.8

-

  1. boot Bootstrap, CP Coverage probability, NDE Natural direct effect, NIE Natural indirect effect, RMSE Root mean squared error, SD Standard deviation, TE Total effect