Causality: models, reasoning and inference

Observational Studies

Research questions that motivate most studies in statistics-based sciences are causal in nature. Economists and social scientists are typically interested in estimating causal effects rather than mere associations between variables (e.g., the effects of training programs on subsequent labor market histories); the same is true for epidemiologists and medical doctors (e.g., is smoking causing lung cancer? what is the effect of pollution on health outcomes?).

This chapter looks at interrelated issues concerning causality, mechanisms, and probability with a focus on epidemiology. I argue there is a tendency in epidemiology, one found in other observational sciences I believe, to try to make formal, abstract inference rules do more work than they can. The demand for mechanisms reflects this tendency, because in the abstract it is ambiguous in multiple ways. Using the Pearl directed acyclic framework (DAG), I show how mechanisms in epidemiology can be unnecessary and how they can be either helpful or essential, depending on whether causal relations or causal effect sizes are being examined. Recent work in epidemiology is finding that traditional stratification analysis can be improved by providing explicit DAGs. However, they are not helpful for dealing with moderating variables and other types of complex causality which can be important epidemiology. 978-0-19-957413-1 04-Mckay-Illari-c04-drv Mckay (Typeset by SPi, Chennai) 71 of 928 November 6, 2010 21:9 OUP UNCORRECTED PROOF -FIRST PROOF, 6/11/2010, SPi

Sociologial Methods and Research, 2019

Theories of causation in philosophy ask what makes causal claims true and establish so called truth conditions allowing one to separate causal from non-causal relationships. We argue that social scientists should be aware of truth conditions of causal claims because they imply which method of causal inference can establish whether a specific claim holds true. A survey of social scientists shows that this is worth emphasizing because many respondents have unclear concepts of causation and link methods to philosophical criteria in an incoherent way. We link five major theories of causation to major small and large-n methods of causal inference to provide clear guidelines to researchers and improve dialogue across methods. While most theories can be linked to more than one method, we argue that structural counterfactual theories are most useful for the social sciences since they require neither social and natural laws nor physical processes to assess causal claims.

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...

1991

Conclusion We have examined only a few of the basic questions about causal inference that result from Reichenbach's two principles. We have not considered what happens when the probability distribution is a mixture of distributions from different causal structures, or how unmeasured common causes can be detected, or what inferences can reliably be drawn about causal relations among unmeasured variables, or the exact advantages that experimental control offers.

Epidemiologic Perspectives & Innovations, 2005

Although epidemiology is necessarily involved with elucidating causal processes, we argue that there is little practical need, having described an epidemiological result, to then explicitly label it as causal (or not). Doing so is a convention which obscures the valuable core work of epidemiology as an important constituent of public health practice. We discuss another approach which emphasizes the public health "use value" of research findings in regard to prediction and intervention independent from explicit metaphysical causal claims. Examples are drawn from smoking and lung cancer, with particular focus on the original 1964 Surgeon General's report on smoking and the new version released in 2004. The intent is to help the epidemiologist focus on the pertinent implications of research, which, from a public health point of view, in large part entails the ability to predict and to intervene. Further discussion will center on the importance of differentiating between technical/practical uses of causal language, as might be used in structural equations or marginal structural modeling, and more foundational notions of cause. We show that statistical/epidemiological results, such as "smoking two packs a day increases risk of lung cancer by 10 times" are in themselves a kind of causal argument that are not in need of additional support from relatively ambiguous language such as "smoking causes lung cancer." We will show that the confusion stemming from the use of this latter statement is more than mere semantics. Our goal is to allow researchers to feel more confident in the power of their research to tell a convincing story without resorting to metaphysical/unsupportable notions of cause.

Bayesian Analysis, 2015

While statisticians and quantitative social scientists typically study the "effects of causes" (EoC), Lawyers and the Courts are more concerned with understanding the "causes of effects" (CoE). EoC can be addressed using experimental design and statistical analysis, but it is less clear how to incorporate statistical or epidemiological evidence into CoE reasoning, as might be required for a case at Law. Some form of counterfactual reasoning, such as the "potential outcomes" approach championed by Rubin, appears unavoidable, but this typically yields "answers" that are sensitive to arbitrary and untestable assumptions. We must therefore recognise that a CoE question simply might not have a well-determined answer. It is nevertheless possible to use statistical data to set bounds within which any answer must lie. With less than perfect data these bounds will themselves be uncertain, leading to a compounding of different kinds of uncertainty. Still further care is required in the presence of possible confounding factors. In addition, even identifying the relevant "counterfactual contrast" may be a matter of Policy as much as of Science. Defining the question is as non-trivial a task as finding a route towards an answer. This paper develops some technical elaborations of these philosophical points, and illustrates them with an analysis of a case study in child protection.

Electronic Workshops in Computing, 2009

Motivation-This paper describes the initial results of a naturalistic inquiry into the way people derive causal inferences. Research approach-We examined media accounts of economic, political, military, and sports incidents to determine the types of causal explanations that are commonly invoked. Findings-We found two interacting processes at work: the identification of potential causes and the framing of these causes into explanations. Explanations took several forms: abstractions, events, lists (undifferentiated collections of partial causes), conditions, and stories (complex mechanisms linking several causes). Originality-Causal reasoning in "the real world" is both different from and far richer than the formal causal accounts found in philosophy, and from the determinate search for causes during scientific problem solving. Takeaway message-By understanding the way causal reasoning is done in natural settings we should be better able to help decision makers diagnose problems and anticipate consequences.

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 ...

Epidemiologic Perspectives & Innovations, 2009

As noted by Wesley Salmon and many others, causal concepts are ubiquitous in every branch of theoretical science, in the practical disciplines and in everyday life. In the theoretical and practical sciences especially, people often base claims about causal relations on applications of statistical methods to data. However, the source and type of data place important constraints on the choice of statistical methods as well as on the warrant attributed to the causal claims based on the use of such methods. For example, much of the data used by people interested in making causal claims come from non-experimental, observational studies in which random allocations to treatment and control groups are not present. Thus, one of the most important problems in the social and health sciences concerns making justified causal inferences using non-experimental, observational data. In this paper, I examine one method of justifying such inferences that is especially widespread in epidemiology and th...