Table 1.
Coefficients and standard errors of the linear association between Y (outcome) and A (exposure) illustrating confounding and collider effects, n = 1000
| Dependent variable (Y) |
|||||
|---|---|---|---|---|---|
| W (confounder) |
C (collider) |
||||
| Unadjusted coefficient (standard error) | Adjusted coefficients (standard error) | Unadjusted coefficient (standard error) | Adjusted coefficients (standard error) | ||
| (Fit 1) | (Fit 2) | (Fit 3) | (Fit 4) | ||
| A | 0.471 | 0.289 | A | 0.326 | −0.416 |
| (0.030) | (0.032) | (0.031) | (0.035) | ||
| W | 0.425 | C | 0.491 | ||
| (0.035) | (0.018) | ||||
| Intercept | −0.061 | −0.060 | 0.010 | 0.035 | |
| (0.033) | (0.031) | (0.031) | (0.023) | ||
| AIC | 100.420 | −31.992 | −55.369 | −626.824 | |
Note: lower AIC is better.