Selection without Collider Stratification can Bias Instrumental Variable Analyses in the Presence of Causal Effect Heterogeneity
- 1University of Bristol
Abstract
Selection bias is a common concern in causal inference. Two mechanisms can give rise to selection bias: collider bias or stratifying on an effect modifier of the exposure-outcome causal effect. Research into the impact of selection bias on instrumental variable (IV) analyses has focused primarily on collider bias. Here, we show that IV estimates can be biased if selection is an effect modifier (or caused by an effect modifier) of the exposure-outcome effect. In a simulation study, we investigate the size and direction of the bias when estimating effects on the mean difference or odds ratio scale. Bias can also occur due to heterogeneity in the instrument-exposure relationship, although its magnitude is typically smaller. Finally, we discuss methods to deal with this bias in practice, including inverse probability weighting (IPW).
Related
- ICONIC: An R Package for Integrating Instrumental Variable- and Negative-Control-Informed Causal Discovery and Diagnostics in Multiomic Studies
- Primary Care Quality and Inappropriate Community Antibiotic Use: A Double Machine Learning Instrumental Variable Approach
- Heterogeneous impact of health insurance on children’s nutritional status: insights from Instrumental Variable Quantile Regression (IV-QR) approach
- Adoption and outcomes of robotic pancreatectomy in the United States: An instrumental variable analysis.
- Radiomics-based machine learning models decipher heterogeneity of anoikis-related molecular subtypes for prognostic stratification in head and neck squamous cell carcinoma
BibTeX
@misc{gkatzionis2026selection,
title = {{Selection without Collider Stratification can Bias Instrumental Variable Analyses in the Presence of Causal Effect Heterogeneity}},
author = {A. Gkatzionis and F. P. Hartwig and K. Tilling},
howpublished = {medRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.27.26364122},
url = {https://www.medrxiv.org/content/10.64898/2026.09.27.26364122v1},
note = {CC BY-NC},
}
RIS
TY - UNPB
TI - Selection without Collider Stratification can Bias Instrumental Variable Analyses in the Presence of Causal Effect Heterogeneity
AU - A. Gkatzionis
AU - F. P. Hartwig
AU - K. Tilling
PY - 2026
DA - 2026/09/27
PB - medRxiv
DO - 10.64898/2026.09.27.26364122
UR - https://www.medrxiv.org/content/10.64898/2026.09.27.26364122v1
AB - Selection bias is a common concern in causal inference. Two mechanisms can give rise to selection bias: collider bias or stratifying on an effect modifier of the exposure-outcome causal effect. Research into the impact of selection bias on instrumental variable (IV) analyses has focused primarily on collider bias. Here, we show that IV estimates can be biased if selection is an effect modifier (or caused by an effect modifier) of the exposure-outcome effect. In a simulation study, we investigate the size and direction of the bias when estimating effects on the mean difference or odds ratio scale. Bias can also occur due to heterogeneity in the instrument-exposure relationship, although its magnitude is typically smaller. Finally, we discuss methods to deal with this bias in practice, including inverse probability weighting (IPW).
ER -