Review of the book “Causal Inference for Statistics, Social, and Biomedical Sciences” by G.W. Imbens and D.B. Rubin

Causal Inference in medicine and in health policy, a summary

2021

A data science task can be deemed as making sense of the data or testing a hypothesis about it. The conclusions inferred from data can greatly guide us to make informative decisions. Big data has enabled us to carry out countless prediction tasks in conjunction with machine learning, such as identifying high risk patients suffering from a certain disease and taking preventable measures. However, healthcare practitioners are not content with mere predictions - they are also interested in the cause-effect relation between input features and clinical outcomes. Understanding such relations will help doctors treat patients and reduce the risk effectively. Causality is typically identified by randomized controlled trials. Often such trials are not feasible when scientists and researchers turn to observational studies and attempt to draw inferences. However, observational studies may also be affected by selection and/or confounding biases that can result in wrong causal conclusions. In thi...

Statistical modeling and causality in social sciences

Institute de Statistique Discussion Paper, 2006

Philosophers and statisticians have been debating on causality for a long time. However, these discussions have been led quite independently from each other. An objective of this paper is to restore a fruitful dialogue between philosophy and statistics. As is well known, at the beginning of the 20th century, some philosophers and statisticians dismissed the concept of causality altogether. It will suffice to mention Bertrand Russell (1913) and Karl Pearson (1911). Almost a hundred years later, causality still represents a central topic ...

Commentary: Estimating causal effects

International Journal of Epidemiology, 2002

Although one goal of aetiologic epidemiology is to estimate 'the true effect' of an exposure on disease occurrence, epidemiologists usually do not precisely specify what 'true effect' they want to estimate. We describe how the counterfactual theory of causation, originally developed in philosophy and statistics, can be adapted to epidemiological studies to provide precise answers to the questions 'What is a cause?', 'How should we measure effects?' and 'What effect measure should epidemiologists estimate in aetiologic studies?' We also show that the theory of counterfactuals (1) provides a general framework for designing and analysing aetiologic studies; (2) shows that we must always depend on a substitution step when estimating effects, and therefore the validity of our estimate will always depend on the validity of the substitution; (3) leads to precise definitions of effect measure, confounding, confounder, and effect-measure modification; and (4) shows why effect measures should be expected to vary across populations whenever the distribution of causal factors varies across the populations.

Reconciling Causality and Statistics

ArXiv, 2020

Statisticians have warned us since the early days of their discipline that experimental correlation between two observations by no means implies the existence of a causal relation. The question about what clues exist in observational data that could informs us about the existence of such causal relations is nevertheless more that legitimate. It lies actually at the root of any scientific endeavor. For decades however the only accepted method among statisticians to elucidate causal relationships was the so called Randomized Controlled Trial. Besides this notorious exception causality questions remained largely taboo for many. One reason for this state of affairs was the lack of an appropriate mathematical framework to formulate such questions in an unambiguous way. Fortunately thinks have changed these last years with the advent of the so called Causality Revolution initiated by Judea Pearl and coworkers. The aim of this pedagogical paper is to present their ideas and methods in a co...

Causal Inference Perspectives

Observational Studies

Extracting information and drawing inferences about causal effects of actions, interventions, treatments and policies is central to decision making in many disciplines and is broadly viewed as causal inference. It was a pleasure to read the lengthy interviews of four leaders in causality and causal inference whose work had such a huge impact on empirical research in many fields. I am honored to follow up on these interviews and share my journey and thoughts of the field and its future.

Response: Defining and estimating causal effects

International Journal of Epidemiology, 2002

Causal Inference in Biomedical Research

Biology and Philosophy

Causation can be inferred by two distinct patterns of reasoning, each requiring a distinct experimental design. Common, non-statistical causal inference is associated with controlled experiments in basic biomedical research. Statistical inference is associated with Randomized Controlled Trials in clinical research. The main difference between the two patterns of inference hinges on the satisfaction of a comparability requirement, which is in turn dictated by the nature of the objects of study, namely homogeneous vs. heterogeneous populations of biological systems. This distinction entails that the objection according to which randomized experiments fail to provide better evidence for causation because randomization cannot guarantee comparability is mistaken. As far as the validity of the statistical inference is concerned, randomization is not required in order to ensure comparability, but rather to prevent systematic bias which may compromise the accuracy of the intervention.

Improving the Validity of Causal Inferences in Observational Studies

AJN, American Journal of Nursing

Previous articles in this series discussed observational study designs, including crosssectional, cohort, and case-control studies. This paper focuses on understanding causal inferences and methods to improve them for observational studies. For example, an observational study might show an association (correlation) between a predictor variable (independent variable) and an outcome (dependent variable); however, the results may not represent cause and effect (Hulley, 2013). Causal inference implies an intervention, e.g., treatment or behavior was the 'cause' of the effect (or outcome). Understanding causal inferences between predictor(s) and outcome(s) can provide insights to understanding the etiology of a disease, identify methods to prevent or reduce disease (or occurrence) and potentially initiate the development of treatments (Hulley, 2013). For example, are eating carrots associated with improved eye health? It is important to note that some associations found in an observational study do not represent cause and effect. However, there are well-recognized explanations for associations between a predictor and outcome in such studies (Hulley, 2013). These occur by chance (random error), bias (systematic error), and confounding variables. The remainder of this paper will describe each item and methods to minimize these factors.

Causation in epidemiology

Journal of Epidemiology and Community Health, 2001

Causation is an essential concept in epidemiology, yet there is no single, clearly articulated definition for the discipline. From a systematic review of the literature, five categories can be delineated: production, necessary and suYcient, suYcient-component, counterfactual, and probabilistic. Strengths and weaknesses of these categories are examined in terms of proposed characteristics of a useful scientific definition of causation: it must be specific enough to distinguish causation from mere correlation, but not so narrow as to eliminate apparent causal phenomena from consideration. Two categoriesproduction and counterfactual-are present in any definition of causation but are not themselves suYcient as definitions. The necessary and suYcient cause definition assumes that all causes are deterministic. The suYcient-component cause definition attempts to explain probabilistic phenomena via unknown component causes. Thus, on both of these views, heavy smoking can be cited as a cause of lung cancer only when the existence of unknown deterministic variables is assumed. The probabilistic definition, however, avoids these assumptions and appears to best fit the characteristics of a useful definition of causation. It is also concluded that the probabilistic definition is consistent with scientific and public health goals of epidemiology. In debates in the literature over these goals, proponents of epidemiology as pure science tend to favour a narrower deterministic notion of causation models while proponents of epidemiology as public health tend to favour a probabilistic view. The authors argue that a single definition of causation for the discipline should be and is consistent with both of these aims. It is concluded that a counterfactually-based probabilistic definition is more amenable to the quantitative tools of epidemiology, is consistent with both deterministic and probabilistic phenomena, and serves equally well for the acquisition and the application of scientific knowledge. (J Epidemiol Community Health 2001;55:905-912) "The view [of causation] we adopt has consequences which reach far beyond informal discussion during coVee breaks." 1 Causation is an essential concept in the practice of epidemiology. Causal claims like "smoking causes cancer" or "human papilloma virus causes cervical cancer" have long been a standard part of the epidemiology literature.

Review of the book “Causal Inference for Statistics, Social, and Biomedical Sciences” by G.W. Imbens and D.B. Rubin (original) (raw)