Cause, Effect, and Everything in Between: An Introduction to Causal Inference, by AboozarHadavand (Oxford, Oxford University Press, 2025), 138 pages + ix front matter

In Cause, Effect, and Everything in Between, Aboozar Hadavand opens with a personal anecdote about his great-grandfather, who reportedly lived to be over one hundred years old. The secret to his longevity? His family is convinced: yogurt, consumed daily. As they saw it, “everyone around him concluded it was the yogurt that made him live so long.” The author, who studied engineering as an undergraduate and economics as a graduate student, uses this relatable story to launch a brief and deliberately non-technical journey through ideas and concepts familiar to anyone with a PhD in economics—namely, how to learn about causal effects from observational studies, quasi-experiments and randomised controlled experiments. Given the title of the book, it may come as a surprise that the term ‘causal effect’ is never rigorously defined. In fairness, Hadavand does allude to Pearl, who says that ‘X is a cause of Y if Y listens to X and decides its value in response to what it hears’. This, of course, is nowhere near a rigorous definition of causality, but the book shows us that one is not needed. The book decidedly sidesteps the sinkhole of defining causality—after all, ask 10 economists about their idea of a causal effect and you will receive 15 different answers. In a way, this book adheres to the mainstream textbook definition, something along the lines of ‘a causal effect is the expected effect of an intervention in a randomized controlled experiment’. The book does not state this explicitly, but in practice, it follows this approach, educating the reader by guiding them through a wealth of examples. To name a few: Do people who own a lot of books earn more? Are people who meditate less anxious? Does drinking alcohol lead to cancer? Is access to health insurance a determinant of physical health? The causal complexities underlying these questions are apparent, and the author does a good job of explaining them to a non-technical audience. The main tools are directed acyclic graphs—a convenient device that avoids the need for equations. The text is unapologetic in its pragmatism. It zips past the long intellectual arc of causality: There is no mention of Thucydides, Aristotle, Spinoza, Hume, Kant, Reichenbach, Cartwright or Lewis (to name a few). Instead, we learn about treatment effects, drug trials, confounding bias, selection bias, regression discontinuity designs, synthetic control methods and difference-in-differences analyses. The approach is ostensibly to shut up and draw a directed acyclic graph. To illustrate: suppose you are interested in the causal effect of melatonin on hours of sleep. In a randomised controlled experiment, you can randomly assign melatonin. The directed acyclic graph looks like this: Coin toss → Melatonin → Hours of sleep. The average causal effect is then defined as the difference in average hours of sleep between those who won and those who lost the coin toss. Using the terminology from the book, this is an example of a ‘forward causal question’: given a cause, how large is its effect on an outcome of interest? In contrast, a ‘reverse causal question’ asks: What are all the causes of hours of sleep? Melatonin is one, but so are alcohol and caffeine consumption, among many others. The book explains well that in randomised controlled experiments, thinking about the reverse causal question is not necessary. If all we are interested in is the causal effect of melatonin on sleep, then randomising its uptake will give us the answer. Yet in the absence of a randomised controlled experiment, thinking about the reverse causal question becomes crucial for making causal claims. The text asks: ‘But how do we come up with causal graphs in the first place?’ I would add and emphasise: How do we come up with complete causal graphs? That is to say, an exhaustive list of variables and their inter-dependencies that allow us to isolate and identify causal effects? The book merely alludes to ‘expert knowledge and intuition’, and I am sympathetic to this dodge. When I teach my students about omitted variable bias and they ask, ‘But how do we know which variables are being omitted?’ (the known unknowns and the unknown unknowns), my standard response involves allusions to ‘best practices’, ‘related research’, ‘experience’ and ‘structural modelling’. It would be unfair to expect more from this brief book. This leaves the question: Who is this book for? I see at least three audiences. First, undergraduate students in statistics, computer science, machine learning or economics will find that this book complements their studies by offering many well-presented applications that untangle the complexities of confounding bias. Second, teachers of these disciplines can use the book as a source of examples to supplement the theoretical aspects of causal analysis with practical narratives. Third, anyone bewildered by the endless barrage of causal claims in everyday life (can I skip that seven-minute workout if I have a glass of red wine with my Mediterranean diet?) will find much to appreciate here. This book provides simple tools to see through false claims of causality and, at the very least, will turn you into an educated sceptic. Readers looking for broader, more detailed or more rigorous treatments could consult Sloman's Causal Models, Pearl's The Book of Why, Pearl, Glymour and Jewell's Causal Inference in Statistics, or Spirtes, Glymour and Scheines' Causation, Prediction and Search. Data sharing not applicable to this article as no datasets were generated or analysed during the current study.