Written by pioneers in the field, this practical book presents an authoritative yet accessible overview of the methods and applications of causal inference. The text provides a thorough introduction to the basics of the theory for non-time-varying treatments and the generalization to complex longitudinal data.

An accessible and contemporary introduction to the methods for determining cause and effect in the social sciences Causal inference encompasses the tools that allow social scientists to determine what causes what. Economists--who generally can't run controlled experiments to test and validate their hypotheses--apply these tools to observational data to make connections. In a messy world, causal inference is what helps establish the causes and effects of the actions being studied, whether the impact (or lack thereof) of (...) increases in the minimum wage on employment, the effects of early childhood education on incarceration later in life, or the introduction of malaria nets in developing regions on economic growth. Scott Cunningham introduces students and practitioners to the methods necessary to arrive at meaningful answers to the questions of causation, using a range of modeling techniques and coding instructions for both the R and Stata programming languages. - -. (shrink)

Causality is central to the understanding and use of data; without an understanding of cause and effect relationships, we cannot use data to answer important questions in medicine and many other fields.

"Correlation is not causation" is one of the mantras of the sciences—a cautionary warning especially to fields like epidemiology and pharmacology where the seduction of compelling correlations naturally leads to causal hypotheses. The standard view from the epistemology of causation is that to tell whether one correlated variable is causing the other, one needs to intervene on the system—the best sort of intervention being a trial that is both randomized and controlled. In this paper, we argue that some purely (...) correlational data contains information that allows us to draw causal inferences: statistical noise. Methods for extracting causal knowledge from noise provide us with an alternative to randomized controlled trials that allows us to reach causal conclusions from purely correlational data. (shrink)

Throughout the Covid-19 pandemic, people have been eager to learn what factors, and especially what public health policies, cause infection rates to wax and wane. But figuring out conclusively what causes what is difficult in complex systems with nonlinear dynamics, such as pandemics. We review some of the challenges that scientists have faced in answering quantitative causal questions during the Covid-19 pandemic, and suggest that these challenges are a reason to augment the moral dimension of conversations about causal (...) inference. We take a lesson from Martha Nussbaum—who cautions us not to think we have just one question on our hands when we have at least two—and apply it to modeling for causal inference in the context of cost-benefit analysis. (shrink)

Over the last two decades, a fundamental outline of a theory of causal inference has emerged. However, this theory does not consider the following problem. Sometimes two or more measured variables are deterministic functions of one another, not deliberately, but because of redundant measurements. In these cases, manipulation of an observed defined variable may actually be an ambiguous description of a manipulation of some underlying variables, although the manipulator does not know that this is the case. In this (...) article we revisit the question of precisely characterizing conditions and assumptions under which reliable inference about the effects of manipulations is possible, even when the possibility of “ambiguous manipulations” is allowed. (shrink)

There has been an intense discussion, albeit largely an implicit one, concerning the inference of causal hypotheses from statistical correlations in quantum mechanics ever since John Bell’s first statement of his notorious theorem in 1966. As is well known, its focus has mainly been the so-called Einstein-Podolsky-Rosen (“EPR”) thought experiment, and the ensuing observed correlations in real EPR like experiments. But although implicitly the discussion goes as far back as Bell’s work, it is only in the last two (...) decades that it has become recognizably and explicitly a debate about causal inference in the quantum realm. The bulk of this paper is devoted to a review of three influential arguments in the philosophical literature that aim to show that causal models for the EPR correlations are impossible, due to Bas Van Fraassen, Daniel Hausman and Huw Price. I contend that all these arguments are inconclusive since they contain premises or presuppositions that are false, unwarranted, or at least controversial. Five different common cause models are outlined that seem perfectly viable for the EPR correlations. These models are then employed to illustrate various difficulties with the premises and presuppositions underlying Van Fraassen’s, Hausman’s and Price’s arguments. In all these cases it is argued that the difficulties cut deep against these authors’ own theories of causation and causal inference. My conclusions are that causal models for the EPR correlations remain viable, that philosophical work is still required to assess their relative virtues, and that in any case the mere theoretical conceivability and empirical possibility of these models sheds doubts over Van Fraassen’s, Hausman’s and (important elements in) Price’s theories of causation and causal inference. (shrink)

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. A good deal is known about these questions, and there (...) is a good deal more to find out. (shrink)

Current debates surrounding the virtues and shortcomings of randomization are symptomatic of a lack of appreciation of the fact that causation can be inferred by two distinct inference methods, each requiring its own, specific experimental design. There is a non-statistical type of inference associated with controlled experiments in basic biomedical research; and a statistical variety associated with randomized controlled trials in clinical research. I argue that the main difference between the two hinges on the satisfaction of the comparability (...) requirement, which is in turn dictated by the nature of the objects of study, namely homogeneous or heterogeneous populations of biological systems. Among other things, this entails that the objection according to which randomized experiments fail to provide better evidence for causation because randomization cannot guarantee comparability is mistaken. (shrink)

Semmelweis’s discovery of the cause of puerperal fever around the middle of the 19th century counts among the paradigm cases of scientific discovery. For several decades, philosophers of science have used the episode to illustrate, appraise and compare views of proper scientific methodology.Here I argue that the episode can be profitably reexamined in light of two cognate notions: causal reasoning and mechanisms. Semmelweis used several causal reasoning strategies both to support his own and to reject competing hypotheses. However, (...) these strategies have gone unappreciated in the existing literature. I show that a causal reasoning approach makes sense of the multitude of tables in Semmelweis’s main text, which in later editions were often abridged because they appeared redundant.Moreover, the existing literature tends to focus on Semmelweis’s clinical intervention and on the extent to which it alone confirms his theoretical conclusions. This neglects Semmelweis’s efforts to show by animal experiments that his clinical results are in agreement with a demonstrable mechanism of puerperal fever pathogenesis. I argue that the full evidential force of Semmelweis’s argument can only be appreciated if both his clinical and his laboratory investigations are taken into account. (shrink)

We argue that a modified version of Mill’s method of agreement can strongly confirm causal generalizations. This mode of causal inference implicates the explanatory virtues of mechanism, analogy, consilience, and simplicity, and we identify it as a species of Inference to the Best Explanation (IBE). Since rational causal inference provides normative guidance, IBE is not a heuristic for Bayesian rationality. We give it an objective Bayesian formalization, one that has no need of principles of (...) indifference and yields responses to the Voltaire objection, van Fraassen’s Bad Lot objection, and John Norton’s recent objection to IBE. (shrink)

Transcranial magnetic stimulation is used to make inferences about relationships between brain areas and their functions because, in contrast to neuroimaging tools, it modulates neuronal activity. The central aim of this article is to critically evaluate to what extent it is possible to draw causal inferences from repetitive TMS data. To that end, we describe the logical limitations of inferences based on rTMS experiments. The presented analysis suggests that rTMS alone does not provide the sort of premises that are (...) sufficient to warrant strong inferences about the direct causal properties of targeted brain structures. Overcoming these limitations demands a close look at the designs of rTMS studies, especially the methodological and theoretical conditions which are necessary for the functional decomposition of the relations between brain areas and cognitive functions. The main points of this article are that TMS-based inferences are limited in that stimulation-related causal effects are not equivalent to structure-related causal effects due to TMS side effects, the electric field distribution, and the sensitivity of neuroimaging and behavioral methods in detecting structure-related effects and disentangling them from confounds. Moreover, the postulated causal effects can be based on indirect effects. A few suggestions on how to manage some of these limitations are presented. We discuss the benefits of combining rTMS with neuroimaging in experimental reasoning and we address the restrictions and requirements of rTMS control conditions. The use of neuroimaging and control conditions allows stronger inferences to be gained, but the strength of the inferences that can be drawn depends on the individual experiment’s designs. Moreover, in some cases, TMS might not be an appropriate method of answering causality-related questions or the hypotheses have to account for the limitations of this technique. We hope this summary and formalization of the reasoning behind rTMS research can be of use not only for scientists and clinicians who intend to interpret rTMS results causally but also for philosophers interested in causal inferences based on brain stimulation research. (shrink)

Whenever the use of non-experimental data for discovering causal relations or predicting the outcomes of experiments or interventions is contemplated, two difficulties are routinely faced. One is the problem of latent variables, or confounders: factors influencing two or more measured variables may not themselves have been measured or recorded. The other is the problem of sample selection bias: values of the variables or features under study may themselves influence whether a unit is included in the data sample.

At the close of his account of causality in the Treatise, Hume acknowledges that he had to adopt the “seemingly preposterous method” of examining the causal inference prior to analyzing the causal relation since the relation “depends so much on the inference”. This dependence emerges in his two definitions of ‘cause’ which, he concedes, seem “extraneous” to the causal relation. In this paper, I try to do what Hume did not do but could have done: (...) fully describe the causal relation so as to reach its core. That core is what I call “existential dependence,” the fact that one “object” exists because another exists. It turns out that Hume’s “preposterous method” is correct since existential dependence is what I call a “distributive” relation, a kind of conjunction that can be represented only by means of certain inferences. This makes existential dependence seem mysterious, but the mystery disappears if we interpret this dependence as probability, something that is comprehensible only through inferences. The key consequence of all of this is that we need to distinguish between the kind of causality that conforms to Hume’s two definitions of ‘cause’ and the kind of causality our causal inferences designate. (shrink)

Philosophical discussions on causal inference in medicine are stuck in dyadic camps, each defending one kind of evidence or method rather than another as best support for causal hypotheses. Whereas Evidence Based Medicine advocates the use of Randomised Controlled Trials and systematic reviews of RCTs as gold standard, philosophers of science emphasise the importance of mechanisms and their distinctive informational contribution to causal inference and assessment. Some have suggested the adoption of a pluralistic approach to (...) causal inference, and an inductive rather than hypothetico-deductive inferential paradigm. However, these proposals deliver no clear guidelines about how such plurality of evidence sources should jointly justify hypotheses of causal associations. We here develop such guidelines by first giving a philosophical analysis of the underpinnings of Hill’s viewpoints on causality. We then put forward an evidence-amalgamation framework adopting a Bayesian net approach to model causal inference in pharmacology for the assessment of harms. Our framework accommodates a number of intuitions already expressed in the literature concerning the EBM vs. pluralist debate on causal inference, evidence hierarchies, causal holism, relevance, and reliability. (shrink)

Previous research suggests that children can infer causal relations from patterns of events. However, what appear to be cases of causal inference may simply reduce to children recognizing relevant associations among events, and responding based on those associations. To examine this claim, in Experiments 1 and 2, children were introduced to a “blicket detector,” a machine that lit up and played music when certain objects were placed upon it. Children observed patterns of contingency between objects and the (...) machine's activation that required them to use indirect evidence to make causal inferences. Critically, associative models either made no predictions, or made incorrect predictions about these inferences. In general, children were able to make these inferences, but some developmental differences between 3- and 4-year-olds were found. We suggest that children's causal inferences are not based on recognizing associations, but rather that children develop a mechanism for Bayesian structure learning. Experiment 3 explicitly tests a prediction of this account. Children were asked to make an inference about ambiguous data based on the base rate of certain events occurring. Four-year-olds, but not 3-year-olds were able to make this inference. (shrink)

Previous research suggests that children can infer causal relations from patterns of events. However, what appear to be cases of causal inference may simply reduce to children recognizing relevant associations among events, and responding based on those associations. To examine this claim, in Experiments 1 and 2, children were introduced to a “blicket detector”, a machine that lit up and played music when certain objects were placed upon it. Children observed patterns of contingency between objects and the (...) machine’s activation that required them to use indirect evidence to make causal inferences. Critically, associative models either made no predictions, or made incorrect predictions about these inferences. In general, children were able to make these inferences, but some developmental differences between 3- and 4- year-olds were found. We suggest that children’s causal inferences are not based on recognizing associations, but rather that children develop a mechanism for Bayesian structure learning. Experiment 3 explicitly tests a prediction of this account. Children were asked to make an inference about ambiguous data based on the base-rate of certain events occurring. Fouryear-olds, but not 3-year-olds were able to make this inference. (shrink)

When a cause interacts with unobserved factors to produce an effect, the contingency between the observed cause and effect cannot be taken at face value to infer causality. Yet, it would be computationally intractable to consider all possible unobserved, interacting factors. Nonetheless, two experiments found that when an unobserved cause is assumed to be fairly stable over time, people can learn about such interactions and adjust their inferences about the causal efficacy of the observed cause. When they observed a (...) period in which a cause and effect were associated followed by a period of the opposite association, rather than concluding a complete lack of causality, subjects inferred an unobserved, interacting cause. The interaction explains why the overall contingency between the cause and effect is low and allows people to still conclude that the cause is efficacious. (shrink)

In medicine and the social sciences, researchers must frequently integrate the findings of many observational studies, which measure overlapping collections of variables. For instance, learning how to prevent obesity requires combining studies that investigate obesity and diet with others that investigate obesity and exercise. Recently developed causal discovery algorithms provide techniques for integrating many studies, but little is known about what can be learned from such algorithms. This article argues that there are causal facts that one could learn (...) by conducting a large study but which could not be learned by combining many smaller studies. Moreover, I characterize the frequency with which combining many studies increases underdetermination and exactly how much information is lost. 1 Introduction2 Causal Inference from Observational Data3 Piecemeal Causal Inference4 The Extent and Frequency of the Problem of Piecemeal Induction5 Conclusion. (shrink)

Instrumental variables allow for quantification of cause and effect relationships even in the absence of interventions. To achieve this, a number of causal assumptions must be met, the most important of which is the independence assumption, which states that the instrument and any confounding factor must be independent. However, if this independence condition is not met, can we still work with imperfect instrumental variables? Imperfect instruments can manifest themselves by violations of the instrumental inequalities that constrain the set of (...) correlations in the scenario. In this article, we establish a quantitative relationship between such violations of instrumental inequalities and the minimal amount of measurement dependence required to explain them for the case of discrete observed variables. As a result, we provide adapted inequalities that are valid in the presence of a relaxed measurement dependence assumption in the instrumental scenario. This allows for the adaptation of existing and new lower bounds on the average causal effect for instrumental scenarios with binary outcomes. Finally, we discuss our findings in the context of quantum mechanics. (shrink)

Causal inference plays a central role in behavioral science. Historically, behavioral science methodologies have typically sought to infer a single causal relation. Each of the major approaches to causal inference in the behavioral sciences follows this pattern. Nonetheless, such approaches sometimes differ in the causal relation that they infer. Incremental causal inference offers an alternative to this conceptualization of causal inference that divides the inference into a series of incremental (...) steps. Different steps infer different causal relations. Incremental causal inference is consistent with both causal pluralism and anti-pluralism. However, anti-pluralism places greater constraints the possible topology of sequential inferences. Arguments against causal inference include questioning consistency with causation as an explanatory principle, charging undue complexity, and questioning the need for it. Arguments in favor of incremental inference include better explanation of diverse causal inferences in behavioral science, tailored causal inference, and more detailed and explicit description of causal inference. Incremental causal inference offers a viable and potentially fruitful alternative to approaches limited to a single causal relation. (shrink)

There are two concepts of causes, property causation and token causation. The principle I want to discuss describes an epistemological connection between the two concepts, which I call the Connecting Principle. The rough idea is that if a token event of type Cis followed by a token event of type E, then the support of the hypothesis that the first event token caused the second increases as the strength of the property causal relation of C to E does. I (...) demonstrate the principle, illustrate its application to phylogenies, infections, and rumours, and discuss its consequence for the conceptual distinctness of causal processes from the events they connect. Although I am by no means confident that the Connecting Principle is ultimately correct, it seems to be a useful point of departure into an important aspect of the epistemology of causality. (shrink)

(2013). Cognitive shortcuts in causal inference. Argument & Computation: Vol. 4, Formal Models of Reasoning in Cognitive Psychology, pp. 64-88. doi: 10.1080/19462166.2012.682655.

Several authors have claimed that mechanisms play a vital role in distinguishing between causation and mere correlation in the social sciences. Such claims are sometimes interpreted to mean that without mechanisms, causal inference in social science is impossible. The author agrees with critics of this proposition but explains how the account of how mechanisms aid causal inference can be interpreted in a way that does not depend on it. Nevertheless, he shows that this more charitable version (...) of the account is still unsuccessful as it stands. Consequently, he advances a proposal for shoring up the account, which is founded on the possibility of acquiring knowledge of social mechanisms by linking together norms or practices found in a society. Key Words: causality • social mechanisms • interpretation • anthropology. (shrink)

This chapter explores the idea that the need to establish common knowledge is one feature that makes social cognition stand apart in important ways from cognition in general. We develop this idea on the background of the claim that social cognition is nothing but a type of causal inference. We focus on autism as our test-case, and propose that a specific type of problem with common knowledge processing is implicated in challenges to social cognition in autism spectrum disorder (...) (ASD). This problem has to do with the individual’s assessment of the reliability of messages that are passed between people as common knowledge emerges. The proposal is developed on the background of our own empirical studies and outlines different ways common knowledge might be comprised. We discuss what these issues may tell us about ASD, about the relation between social and non-social cognition, about social objects, and about the dynamics of social networks. (shrink)

Most causal discovery algorithms in the literature exploit an assumption usually referred to as the Causal Faithfulness or Stability Condition. In this paper, we highlight two components of the condition used in constraint-based algorithms, which we call “Adjacency-Faithfulness” and “Orientation- Faithfulness.” We point out that assuming Adjacency-Faithfulness is true, it is possible to test the validity of Orientation- Faithfulness. Motivated by this observation, we explore the consequence of making only the Adjacency-Faithfulness assumption. We show that the familiar PC (...) algorithm has to be modified to be correct under the weaker, Adjacency-Faithfulness assumption. The modified algorithm, called Conservative PC (CPC), checks whether Orientation- Faithfulness holds in the orientation phase, and if not, avoids drawing certain causal conclusions the PC algorithm would draw. Howtion: ever, if the stronger, standard causal Faith-. (shrink)

For nearly as long as the word “correlation” has been part of statistical parlance, students have been warned that correlation does not prove causation, and that only experimental studies, e.g., randomized clinical trials, can establish the existence of a causal relationship. Over the last few decades, somewhat of a consensus has emerged between statisticians, computer scientists, and philosophers on how to represent causal claims and connect them to probabilistic relations. One strand of this work studies the conditions under (...) which evidence accumulated from non-experimental (observational) studies can be used to infer a causal relationship. In this paper, I compare the typical conditions required to infer that one variable is a direct cause of another in observational and experimental studies. I argue that they are essentially the same. (shrink)

Much of the recent work on the epistemology of causation has centered on two assumptions, known as the Causal Markov Condition and the Causal Faithfulness Condition. Philosophical discussions of the latter condition have exhibited situations in which it is likely to fail. This paper studies the Causal Faithfulness Condition as a conjunction of weaker conditions. We show that some of the weaker conjuncts can be empirically tested, and hence do not have to be assumed a priori. Our (...) results lead to two methodologically significant observations: (1) some common types of counterexamples to the Faithfulness condition constitute objections only to the empirically testable part of the condition; and (2) some common defenses of the Faithfulness condition do not provide justification or evidence for the testable parts of the condition. It is thus worthwhile to study the possibility of reliable causal inference under weaker Faithfulness conditions. As it turns out, the modification needed to make standard procedures work under a weaker version of the Faithfulness condition also has the practical effect of making them more robust when the standard Faithfulness condition actually holds. This, we argue, is related to the possibility of controlling error probabilities with finite sample size (“uniform consistency”) in causal inference. (shrink)

As is well known, the Humean account of causal inference gives a central location to inference habits. Some of these habits one can discipline. Thus, one can so discipline oneself as to reason in accordance with the “rules by which to judge of causes and effects”, that is, one can discipline oneself to think scientifically, rather than, say, in accordance with the rules of prejudice, or of superstition. All such judgments, even those of science, are, however, upon (...) the Humean account fallible. Some, or rather very many, have raised the issue about how any such causal judgment or inference is justified. That is, how can such inference habits be justified, shown to be reasonable. Hume meets this challenge by arguing that there are some habits all men do and must have. Because we must have them we ought to have them. They are, therefore, justified. (shrink)

Frank Cabrera argues that informational explanatory virtues—specifically, mechanism, precision, and explanatory scope—cannot be confirmational virtues, since hypotheses that possess them must have a lower probability than less virtuous, entailed hypotheses. We argue against Cabrera’s characterization of confirmational virtue and for an alternative on which the informational virtues clearly are confirmational virtues. Our illustration of their confirmational virtuousness appeals to aspects of causal inference, suggesting that causal inference has a role for the explanatory virtues. We briefly explore (...) this possibility, delineating a path from Mill’s method of agreement to Inference to the Best Explanation (IBE). (shrink)

S There is a long tradition of representing causal relationships by directed acyclic graphs (Wright, 1934 ). Spirtes ( 1994), Spirtes et al. ( 1993) and Pearl & Verma ( 1991) describe procedures for inferring the presence or absence of causal arrows in the graph even if there might be unobserved confounding variables, and/or an unknown time order, and that under weak conditions, for certain combinations of directed acyclic graphs and probability distributions, are asymptotically, in sample size, consistent. (...) These results are surprising since they seem to contradict the standard statistical wisdom that consistent estimators of causal effects do not exist for nonrandomised studies if there are potentially unobserved confounding variables. We resolve the apparent incompatibility of these views by closely examining the asymptotic properties of these causal inference procedures. We show that the asymptotically consistent procedures are ‘pointwise consistent’, but ‘uniformly consistent’ tests do not exist. Thus, no finite sample size can ever be guaranteed to approximate the asymptotic results. We also show the nonexistence of valid, consistent confidence intervals for causal effects and the nonexistence of uniformly consistent point estimators. Our results make no assumption about the form of the tests or estimators. In particular, the tests could be classical independence tests, they could be Bayes tests or they could be tests based on scoring methods such as  or . The implications of our results for observational studies are controversial and are discussed briefly in the last section of the paper. The results hinge on the following fact: it is possible to find, for each sample size n, distributions P and Q such that P and Q are empirically indistinguishable and yet P and Q correspond to different causal effects. (shrink)

The literature on causal discovery has focused on interventions that involve randomly assigning values to a single variable. But such a randomized intervention is not the only possibility, nor is it always optimal. In some cases it is impossible or it would be unethical to perform such an intervention. We provide an account of ‘hard' and ‘soft' interventions and discuss what they can contribute to causal discovery. We also describe how the choice of the optimal intervention(s) depends heavily (...) on the particular experimental setup and the assumptions that can be made. ‡The first author is funded by the Causal Learning Collaborative Initiative supported by the James S. McDonnell Foundation. Many aspects of this paper were inspired by discussions with members of the collaborative. †To contact the authors, please write to: Department of Philosophy, Carnegie Mellon University, Pittsburgh, PA 15213; e-mail: [email protected] and [email protected]. (shrink)

In Causation, Prediction, and Search (CPS hereafter), Peter Spirtes, Clark Glymour and I developed a theory of statistical causal inference. In his presentation at the Notre Dame conference (and in his paper, this volume), Glymour discussed the assumptions on which this theory is built, traced some of the mathematical consequences of the assumptions, and pointed to situations in which the assumptions might fail. Nevertheless, many at Notre Dame found the theory difficult to understand and/or assess. As a result (...) I was asked to write this paper to provide a more intuitive introduction to the theory. In what follows I shun almost all formality and avoid the numerous and complicated qualifiers that typically accompany definitions or important philosophical concepts. They can be all be found in Glymour's paper or in CPS, which are clear although sometimes dense. Here I attempt to fix intuitions by highlighting a few of the essential ideas and by providing extremely simple examples throughout. (shrink)

Why do classic biostatistical studies, alleged to provide causal explanations of effects, often fail? This article argues that in statistics-relevant areas of biology—such as epidemiology, population biology, toxicology, and vector ecology—scientists often misunderstand epistemic constraints on use of the statistical-significance rule (SSR). As a result, biologists often make faulty causal inferences. The paper (1) provides several examples of faulty causal inferences that rely on tests of statistical significance; (2) uncovers the flawed theoretical assumptions, especially those related to (...) randomization, that likely contribute to flawed biostatistics; (3) re-assesses the three classic (SSR-warrant, avoiding-selection-bias, and avoiding-confounders) arguments for using SSR only with randomization; and (4) offers five new reasons for biologists to use SSR only with randomized experiments. (shrink)

The manipulationist account of causation provides a conceptual analysis of cause-effect relationships in terms of hypothetical experiments. It also explains why and how experiments are used for the empirical testing of causal claims. This paper attempts to apply the manipulationist account of causation to a broader range of experiments—a range that extends beyond experiments explicitly designed for the testing of causal claims. I aim to show that the set of causal inferences afforded by an experiment is determined (...) solely on the basis of contrasting case structures that I call “experimental series”, and that the conditions that suffice for causal inference obtain quite commonly, even among “ordinary” experiments that are not explicitly designed for the testing of causal claims. (shrink)

One conception of underdetermination is that it corresponds to the impossibility of reliable inquiry. In other words, underdetermination is defined to be the situation where, given a set of background assumptions and a space of hypotheses, it is logically impossible for any hypothesis selection method to meet a given reliability standard. From this perspective, underdetermination in a given subject of inquiry is a matter of interplay between background assumptions and reliability or success criteria. In this paper I discuss underdetermination in (...) causal inference along this line. In particular I will analyze several success criteria that can be applied to causal inference from statistical regularities. The criteria center on the notions of consistency in mathematical statistics. For each criterion I present its epistemic implication in terms of simple conditions under which the criterion cannot possibly be met. I then investigate which of the familiar principles and their variants in the literature, if adopted as background assumptions, are sufficient to overcome different levels of underdetermination induced by the success criteria. “亚决定性”是知识论和科学哲学中一个重要的概念。对这个概念的一种阐释是把它对应于可靠探索的不可能性。就是说，在一个（经验）问题中，给定一些公设和一些供选择的理论或假说，如果逻辑上不可能找到一种理论选择 的方法能满足一定的可靠或成功标准，那么相对于这个标准就存在亚决定性。从这个观点看，亚决定性总是相对于一个问题设定，尤其是公设和成功标准而言的。本文从这个角度对近来的统计因果推理研究作一番梳理。首先，基 于数理统计中的一致性概念，我会讨论和分析一系列可应用于因果推理的成功标准。对每一个标准，我会用一个相对简单的条件来刻画它对应的亚决定性。然后我对文献里一部分重要的结果作一个综述，以澄清什么样的公设可以 消除什么样的亚决定性。. (shrink)

We hear that a glass of red wine prolongs life, that alcohol is a carcinogen, that pregnant women should drink not a drop of alcohol. Major medical journals first claimed that hormone replacement therapy reduces the risk of heart disease, then reversed themselves and said it increases the risk of heart disease. What are the effects caused by consuming alcohol or by receiving hormone replacement therapy? These are causal questions, questions about the effects caused by treatments, policies or preventable (...) exposures. Some causal questions can be studied in randomized trials, in which a coin is flipped to decide the treatment for the next experimental subject. Because randomized trials are not always practical, nor always ethical, many causal questions are investigated in non-randomized observational studies. The reversal of opinion about hormone replacement therapy occurred when a randomized clinical trial contradicted a series of earlier observational studies. Using minimal mathematics - high school algebra and coin flips -- and numerous examples, Observation and Experiment explains the key concepts and methods of causal inference. Examples of randomized experiments and observational studies are drawn from clinical medicine, economics, public health and epidemiology, clinical psychology and psychiatry. (shrink)

The study of causal inference has seen recent momentum in machine learning and artificial intelligence, particularly in the domains of transfer learning, reinforcement learning, automated diagnostics, and explainability. Yet, despite its increasing application to address many of the boundaries in modern AI, causal topics remain absent in most AI curricula. This work seeks to bridge this gap by providing classroom-ready introductions that integrate into traditional topics in AI, suggests intuitive graphical tools for the application to both new (...) and traditional lessons in probabilistic and causal reasoning, and presents avenues for instructors to impress the merit of climbing the “causal hierarchy” to address problems at the levels of associational, interventional, and counterfactual inference. Finally, this study shares anecdotal instructor experiences, successes, and challenges integrating these lessons at multiple levels of education. (shrink)