Quantifying Biases in Causal Models: Classical Confounding vs Collider-Stratification Bias

By Sander Greenland. 2003

838 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Statistics and Probability

Citations in the corpus

Handling time varying confounding in observational research (2017, background): “Secondly, selection (also known as collider stratification) bias, 11 which occurs by inappropriately adjusting for a time varying confounder that may share a common cause with the outcome.”
Single World Intervention Graphs ( SWIGs ) : A Unification of the Counterfactual and Graphical Approaches to Causality (2013, methodology): “1 in Pearl (2009), shown in Figure 16, is a familiar example of ‘m-bias’ (Greenland, 2003) where Y (a) ?? A, yet Y (a) 6 ??A | Z.”
How to control confounding effects by statistical analysis (2012, background): “A successful randomization minimizes confounding by unmeasured as well as measured factors, whereas statistical control that addresses confounding by measurement and can introduce confounding through inappropriate control (7-9).”
The table 2 fallacy: presenting and interpreting confounder and modifier coefficients. (2013, background): “We note that in Figure 2, the confounding of age modification of the HIV effect measure by U would be entirely due to conditioning on smoking (which is a collider for U and age (17)); again, however, conditioning on smoking was necessary to remove confounding by smoking and by U of the HIV effect me”
Machine learning and deep learning—A review for ecologists (2022, neutral): “An example is severe feature col-linearity, where including all features increases the uncertainties of the estimates (Greenland, 2003; Lederer et al., 2018 ), whereas removing features, even if causally connected to the response, can improve predictive performance (Arif & MacNeil, 2022 ; Dormann…”
Constructing inverse probability weights for marginal structural models. (2008, background): “First, the addition of a nonconfounding variable may introduce selection bias due to collider stratification (29, 30).”