CoChat
Sign Up
Cited by 810Advanced Causal Inference TechniquesStatistical Methods in EpidemiologyBayesian Modeling and Causal Inference

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

Greenland, Sander·2003·Epidemiology
DOI 10.1097/01.ede.0000042804.12056.6c
Open this paper in CoChat — free account

Abstract

It has long been known that stratifying on variables affected by the study exposure can create selection bias. More recently it has been shown that stratifying on a variable that precedes exposure and disease can induce confounding, even if there is no confounding in the unstratified (crude) estimate. This paper examines the relative magnitudes of these biases under some simple causal models in which the stratification variable is graphically depicted as a collider (a variable directly affected by two or more other variables in the graph). The results suggest that bias from stratifying on variables affected by exposure and disease may often be comparable in size with bias from classical confounding (bias from failing to stratify on a common cause of exposure and disease), whereas other biases from collider stratification may tend to be much smaller.

Integrity signals

Registry

Crossref · 10.1097/01.ede.0000042804.12056.6c

Cites

7 of 7 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Claims the paper makes

This paper’s full text isn’t openly available, so its claims and self-check weren’t run. The registry and citation checks still apply.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • This bias can arise from conditioning on the outcome, a post-outcome variable, a post-treatment variable, or a pre-treatment variable.

    “Endogenous selection can result from conditioning on an outcome, but it can also result from conditioning on variables that have been affected by the outcome, variables that temporally precede the outcome, and even variables that temporally precede the treatment (Greenland 2003).”

    From Endogenous Selection Bias: The Problem of Conditioning on a Collider Variable (2014) · cites this paper as Greenland 2003

  • Conditioning on a pre-treatment collider will usually lead to a small bias.

    “Finally, conditioning on a pre-treatment collider (as in Figures 12 and 13) will usually lead to a small bias, which is why recognizing latent homophily bias as endogenous selection bias (as opposed to confounding bias) should be good news to network analysts.”

    From Endogenous Selection Bias: The Problem of Conditioning on a Collider Variable (2014) · cites this paper as Greenland 2003

  • Greenland derived analytic results and approximations for the relative size of endogenous selection bias and confounding bias in DAGs with binary variables.

    “Greenland (2003), in particular, has derived useful analytic results and approximations for the relative size of endogenous selection bias and confounding bias in DAGs with binary variables (see also Kaufman et al. 2009).”

    From Endogenous Selection Bias: The Problem of Conditioning on a Collider Variable (2014) · cites this paper as Greenland 2003

  • If a pre-treatment variable is both a collider and a confounder, conditioning on it may remove more bias than it induces, even though nonparametric point identification is still not feasible.

    “Furthermore, if the pre-treatment variable is both a collider and a confounder (as in Figure 13c), then conditioning on it may remove more bias than it induces, even though nonparametric point identification is still not feasible.”

    From Endogenous Selection Bias: The Problem of Conditioning on a Collider Variable (2014) · cites this paper as Greenland 2003

  • When all variables in the DAG are binary, Greenland suggests that controlling for a variable that is both a confounder and a collider likely but not necessarily removes more bias than it creates.

    “When all variables in the DAG are binary, Greenland (2003) suggests that controlling for a variable that is both a confounder and a collider, as in Figure 13c, likely (but not necessarily) removes more bias than it creates.”

    From Endogenous Selection Bias: The Problem of Conditioning on a Collider Variable (2014) · cites this paper as Greenland 2003

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Authors
Greenland, S.
Affiliations
University of California, Los Angeles (US)
Corresponding author
Greenland, Sander
ORCID
Greenland, Sander
Venue
Epidemiology
Publisher
Ovid Technologies (Wolters Kluwer Health)
Published
2003-05
Volume
14
Issue
3
Pages
300-306
ISSN
  • 1044-3983 (ISSN-L)
  • 1531-5487
DOI
10.1097/01.ede.0000042804.12056.6c
OpenAlex
W1992882281
Cited by
810
References
7
Citations by year
2026: 15 · 2025: 48 · 2024: 47 · 2023: 59 · 2022: 51 · 2021: 64 · 2020: 64 · 2019: 58 · 2018: 39 · 2017: 52 · 2016: 46 · 2015: 31 · 2014: 47 · 2013: 42 · 2012: 27
FWCI
5.49
Type
journal article
Language
en
Open access
closed
Locations
  • Epidemiology (publishedVersion)
  • PubMed (publishedVersion)
Funders
Eunice Kennedy Shriver National Institute of Child Health and Human Development (R01HD039746)
Topics
  • Physical Sciences › Mathematics › Statistics and Probability › Advanced Causal Inference Techniques
  • Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods in Epidemiology
  • Physical Sciences › Computer Science › Artificial Intelligence › Bayesian Modeling and Causal Inference
Keywords
collider stratification bias, selection bias, stratification, particle colliders, causal models, bias quantification, common cause
MeSH terms
Humans, Bias, Causality, Confounding Factors, Epidemiologic

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review