Skip to main content

Table 3 Type I Error Rates for Non-inferiority Test in Simulation Study

From: Modeling observations with a detection limit using a truncated normal distribution with censoring

μ1

μ2

δ

σ

GS1

Uncens NT2

DL3

1/2 DL

Tobit4

tcensReg5

1.1

0.95

-0.15

0.40

0.0552

0.0609

0.1240

0.0477

0.0576

0.0555

1.0

0.85

-0.15

0.40

0.0563

0.0709

0.1874

0.0577

0.0652

0.0594

1.1

0.95

-0.15

0.45

0.0606

0.0736

0.1424

0.0648

0.0712

0.0642

1.0

0.85

-0.15

0.45

0.0642

0.0863

0.2046

0.0766

0.0776

0.0682

1.1

0.95

-0.15

0.50

0.0582

0.0755

0.1456

0.0687

0.0670

0.0599

1.0

0.85

-0.15

0.50

0.0603

0.0887

0.1996

0.0831

0.0754

0.0641

  1. Note: nominal α set to 0.05
  2. 1GS = Gold Standard, i.e., uncensored observations with truncation adjustment
  3. 2Uncens NT = Uncensored data with no truncation adjustment
  4. 3DL = detection limit
  5. 4Tobit = Tobit censored regression with no truncation adjustment
  6. 5tcensReg = Censored regression with truncation adjustment