Article
Statistics & Probability
Gonzalo Vazquez-Bare
Summary: This study proposes a potential outcomes framework to analyze spillover effects using instrumental variables. The findings suggest that intention-to-treat (ITT) parameters do not have a clear link to causally interpretable parameters, and rescaling them by first-stage estimands recovers a weighted combination of average effects. The study also analyzes the identification of causal effects under one-sided noncompliance and introduces an alternative assumption for identifying parameters of interest under two-sided noncompliance.
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION
(2023)
Article
Public, Environmental & Occupational Health
Fernando Pires Hartwig, Linbo Wang, George Davey Smith, Neil Martin Davies
Summary: Instrumental variables (IVs) can be used to determine the causal effect of a treatment X on an outcome Y. Further assumptions, such as homogeneity in the causal effect of X on Y and no effect modification, are needed to identify the average causal effect (ACE) of X on Y. The assumption of no simultaneous heterogeneity is sufficient for identifying the ACE using IVs, even if other assumptions are violated.
Article
Statistics & Probability
Zhichao Jiang, Peng Ding
Summary: Principal stratification is a framework to deal with posttreatment intermediate variables between treatment and outcomes, but the principal causal effects are not identifiable without additional assumptions due to unobservability. Previous studies have shown that using auxiliary variables can improve the inference of principal causal effects.
STATISTICAL SCIENCE
(2021)
Article
Computer Science, Artificial Intelligence
Debo Cheng, Jiuyong Li, Lin Liu, Kui Yu, Thuc Duy Le, Jixue Liu
Summary: Instrumental variable (IV) is a powerful method for inferring causal effects from observational data. However, the selection of a valid IV is crucial as an invalid IV can result in biased estimates. In this article, a data-driven algorithm is proposed to discover valid IVs based on partial ancestral graphs (PAGs). Experiments on synthetic and real-world datasets demonstrate that the algorithm provides accurate estimates of causal effects compared to state-of-the-art IV-based estimators.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Review
Health Care Sciences & Services
Roy S. D. Zawadzki, Joshua D. L. Grill, Daniel L. Gillen
Summary: In order to estimate causal effects, analysts conducting observational studies in health settings use different strategies to reduce bias caused by confounding by indication. These strategies include using confounders and instrumental variables (IVs). As these methods rely on untestable assumptions, analysts must acknowledge that they will perform imperfectly. This tutorial proposes general principles and heuristics for estimating causal effects in both approaches when assumptions are potentially violated.
BMC MEDICAL RESEARCH METHODOLOGY
(2023)
Article
Economics
Michela Bia, Alessandra Mattei, Andrea Mercatanti
Summary: This study addresses important statistical issues in evaluating the effects of training programs for unemployed individuals by proposing an extended framework to simultaneously tackle truncation and attrition issues. The framework utilizes principal stratification to define causal effects and handle missing data, in addition to applying a Bayesian approach for inference. The findings from evaluating the causal effects of foreign language training programs in Luxembourg may provide insights for improving future employment programs.
JOURNAL OF BUSINESS & ECONOMIC STATISTICS
(2022)
Article
Psychology, Clinical
Ellicott C. Matthay, Meghan L. Smith, M. Maria Glymour, Justin S. White, Jaimie L. Gradus
Summary: This article introduces the conceptual foundations, implementation, and strengths and limitations of instrumental variables (IVs) in research. IV methods offer an alternative approach to controlling for confounders without the need for their correct identification, measurement, and control. They are particularly relevant for assessing the causal effects of stress and trauma on outcomes.
PSYCHOLOGICAL TRAUMA-THEORY RESEARCH PRACTICE AND POLICY
(2023)
Article
Statistics & Probability
Frank Windmeijer, Xiaoran Liang, Fernando P. Hartwig, Jack Bowden
Summary: The CI method proposed in this study is based on confidence intervals to select valid instruments from a larger set, ensuring a monotonically decreasing number of instruments selected with decreasing tuning parameter values. The method employs a downward testing procedure with the Sargan test as a major verification step.
JOURNAL OF THE ROYAL STATISTICAL SOCIETY SERIES B-STATISTICAL METHODOLOGY
(2021)
Article
Statistics & Probability
Silvia Noirjean, Mario Biggeri, Laura Forastiere, Fabrizia Mealli, Maria Nannini
Summary: When treatment cannot be enforced but only encouraged, noncompliance is likely to occur. Instrumental Variables methods are commonly used in applied economics to address noncompliance. These methods allow for the identification of causal effects for individuals who comply with the treatment. One important assumption in this identification process is the Exclusion Restriction, which rules out causal effects for individuals who never take the treatment.
STATISTICAL METHODS AND APPLICATIONS
(2023)
Article
Mathematical & Computational Biology
Booil Jo
Summary: This study explores the use of parametric mixture modeling in principal stratification modeling and proposes a method to assess the quality of estimation results. By employing the flexible moving exclusion restriction assumption, the study demonstrates the potential of using parametric mixture modeling as a tool for causal inference.
STATISTICS IN MEDICINE
(2022)
Article
Physics, Multidisciplinary
Feng Xie, Yangbo He, Zhi Geng, Zhengming Chen, Ru Hou, Kun Zhang
Summary: This paper investigates the problem of selecting instrumental variables from observational data generated by linear non-Gaussian acyclic causal models. A necessary condition for detecting variables that cannot serve as instrumental variables is proposed. The graphical implications of this condition are characterized in linear non-Gaussian acyclic causal models. A method to select the set of candidate instrumental variables is developed and experimental results show its effectiveness.
Article
Medicine, General & Internal
Sonja N. Tang, Verena Zuber, Konstantinos K. Tsilidis
Summary: This study used Mendelian randomisation analysis to investigate the causal relationships between biochemical biomarkers in the UK Biobank and breast cancer. The results suggest that genetically predicted levels of testosterone, HDL cholesterol, and IGF-1 may play a causal role in breast cancer development, and a potential novel role of alkaline phosphatase in breast cancer etiology was discovered.
Article
Statistics & Probability
Wing Hung Wong
Summary: When the causal relationship between X and Y is specified by a structural equation, the average causal effect of X on Y can be identified as the solution of an integral equation based on the distributions of (X, Z) and (Y, Z), where Z serves as an instrumental variable. This parameter cannot be determined from the distribution of (X, Y) itself.
Article
Mathematical & Computational Biology
Yanxun Xu, Daniel Scharfstein, Peter Muller, Michael Daniels
Summary: We propose a Bayesian nonparametric approach to evaluate the causal effect of treatment in a randomized trial with semi-competing risks. Our method introduces a novel estimand based on principal stratification and utilizes identification assumptions indexed by a sensitivity parameter.
Article
Public, Environmental & Occupational Health
Haodong Tian, Stephen Burgess
Summary: Multivariable Mendelian randomization (MVMR) is an approach that can assess the effects of related risk factors using genetic variants as instrumental variables, but performs poorly in studying time-varying causal effects, possibly due to model misspecification and violation of the exclusion restriction assumption.
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
(2023)
Article
History & Philosophy Of Science
Philip Dawid
Letter
Medicine, Legal
Geoffrey Stewart Morrison, David H. Kaye, David J. Balding, Duncan Taylor, Philip Dawid, Colin G. G. Aitken, Simone Gittelson, Grzegorz Zadora, Bernard Robertson, Sheila Willis, Susan Pope, Martin Neil, Kristy A. Martire, Amanda Hepler, Richard D. Gill, Allan Jamieson, Jacob de Zoete, R. Brent Ostrum, Amke Caliebe
FORENSIC SCIENCE INTERNATIONAL
(2017)
Article
Social Sciences, Mathematical Methods
David M. Phillippo, Sofia Dias, A. E. Ades, Vanessa Didelez, Nicky J. Welton
JOURNAL OF THE ROYAL STATISTICAL SOCIETY SERIES A-STATISTICS IN SOCIETY
(2018)
Editorial Material
Engineering, Industrial
Willis A. Jensen, Douglas C. Montgomery, Fugee Tsung, G. Geoffrey Vining
JOURNAL OF QUALITY TECHNOLOGY
(2018)
Article
Statistics & Probability
Stijn Vansteelandt, Vanessa Didelez
SCANDINAVIAN JOURNAL OF STATISTICS
(2018)
Article
Mathematics, Interdisciplinary Applications
Vanessa Didelez
LIFETIME DATA ANALYSIS
(2019)
Article
Mathematical & Computational Biology
Janine Witte, Vanessa Didelez
BIOMETRICAL JOURNAL
(2019)
Article
Mathematical & Computational Biology
Odd O. Aalen, Mats J. Stensrud, Vanessa Didelez, Rhian Daniel, Kjetil Roysland, Susanne Strohmaier
BIOMETRICAL JOURNAL
(2020)
Article
Genetics & Heredity
Nuala A. Sheehan, Vanessa Didelez
Editorial Material
Mathematical & Computational Biology
Vanessa Didelez, Mats Julius Stensrud
BIOMETRICAL JOURNAL
(2022)
Article
Biology
Isabel R. Fulcher, Ilya Shpitser, Vanessa Didelez, Kali Zhou, Daniel O. Scharfstein
Summary: A method proposed by Huang for assessing the impact of a point treatment on mortality involves using a causal mediation framework and statistical assumptions for identification, which may be difficult to interpret and justify in some cases.
Article
Social Sciences, Mathematical Methods
Philip Dawid, Macartan Humphreys, Monica Musio
Summary: This study investigates the probability of causation under certain assumptions and finds that even with complete mediators, positive evidence has limited contribution to assessing the probability of causation.
SOCIOLOGICAL METHODS & RESEARCH
(2022)
Article
Health Care Sciences & Services
Malte Braitmaier, Sarina Schwarz, Bianca Kollhorst, Carlo Senore, Vanessa Didelez, Ulrike Haug
Summary: This study evaluated the effectiveness of screening colonoscopy in reducing the incidence of distal and proximal colorectal cancer in individuals aged 55-69 years. The results showed that colonoscopy screening was effective in preventing both distal and proximal CRC. Unlike previous studies, this research utilized a target trial approach and found no relevant difference in effectiveness by location.
JOURNAL OF CLINICAL EPIDEMIOLOGY
(2022)
Article
Health Care Sciences & Services
Elinor Curnow, James R. Capenter, Jon E. Heron, Rosie P. Cornish, Stefan Rach, Vanessa Didelez, Malte Langeheine, Kate Tilling
Summary: Epidemiological studies often have missing data, which are commonly handled by multiple imputation (MI). This study examines the bias caused by the default option of using simple linear covariate functions in the imputation model and provides practical guidance for researchers. The results show that mis-specification of the relationship between outcome and exposure, or between exposure and confounder, can cause bias in MI estimates, and the method of predictive mean matching can mitigate model mis-specification.
JOURNAL OF CLINICAL EPIDEMIOLOGY
(2023)
Editorial Material
Social Sciences, Mathematical Methods
A. Philip Dawid
JOURNAL OF THE ROYAL STATISTICAL SOCIETY SERIES A-STATISTICS IN SOCIETY
(2023)