Skip to main content

Table 3 \(\hat{\beta }_{X|C,U(\phi )}\) [\(95\%\) confidence interval] when \(\phi\) equals multiples of benchmark values for maternal weight (MW)

From: Quantitative bias analysis in practice: review of software for regression with unmeasured confounding

 

\(\hat{\beta }_{X|C,U(\phi )}\ [95\%\) confidence interval] in \(kg/m^2\)

If \(\phi\) set to

treatSens

causalsens

sensemakr

konfound

Bias towards the null

    

   \(0.5 \times\) benchmark values of MW

2.19 [1.29, 3.09]

0.52 \([-0.39, 1.42]\)

2.08 [1.18, 2.97]

[excludes 0]

   \(1 \times\) benchmark values of MW

2.14 [1.24, 3.04]

\(-0.19\) \([-1.09, 0.72]\)  

1.94 [1.06, 2.82]

[excludes 0]

   \(2 \times\) benchmark values of MW

1.91 [1.04, 2.78]

\(-1.18\) \([-2.08, -0.28]\)

1.67 [0.82, 252]

[excludes 0]

Bias away from the null

    

   \(0.5 \times\) benchmark values of MW

2.24 [1.34, 3.14]

3.90 [3.00, 4.80]

2.34 [1.45, 3.23]

[excludes 0]

   \(1 \times\) benchmark values of MW

2.31 [1.42, 3.21]

4.60 [3.70, 5.51]

2.47 [1.60, 3.35]

[excludes 0]

   \(2 \times\) benchmark values of MW

2.54 [1.67, 3.41]

5.59 [4.69, 6.50]

2.74 [1.89, 3.59]

[excludes 0]