Resources
Core Textbooks
As mentioned on the syllabus, the field of causal inference in general (and causal inference for computational social science in particular) moves excitingly-fast, such that the material has yet to “congeal” into a single, all-encompassing textbook. Nonetheless, the following three books cover a substantial amount of ground (described in more detail below each citation), so that together they form a coherent “three-volume textbook” for this class! If you can only read three books this summer, read these :)
Morgan and Winship, Counterfactuals and Causal Inference: Methods and Principles for Social Research [PDF]
This is the book which comes closest to being an all-encompassing, single textbook for the class! It brings together different “strands” of causal modeling research (since each field—economics, bioinformatics, sociology, etc.—tends to use its own notation and vocabulary), unifying them into a single approach. The only reason we can’t use it as the main textbook is because it hasn’t been updated since 2015, and most of the assignments in this class use computational tools from 2018 onwards!
Angrist and Pischke, Mastering ’Metrics: The Path from Cause to Effect (Angrist and Pischke 2014) [PDF]
This book is included as the second of the three “core” texts mainly because, it uses the language of causality specific to Econometrics, the language that is most familiar to me from my PhD training in Political Economy. However, if you tend to learn better by example, it also does a good job of foregrounding specific examples (like evaluating charter schools and community policing policies), so that the methods emerging naturally from attempts to solve these puzzles when association methods like linear regression fail to capture their causal linkages.
Pearl and Mackenzie, The Book of Why: The New Science of Cause and Effect [EPUB]
This book contrasts with the Angrist and Pischke book in using the language of causality formed within Computer Science rather than Economics. It can be a good starting point especially if you’re unfamiliar with the heavy use of diagrams for scientific modeling—basically, whereas Angrist and Pischke’s first instinct is to use (sometimes informal) equations like \(y = mx + b\) to explain steps in the procedures, Pearl and Mackenzie’s instinct would be to instead use something like \(\require{enclose}\enclose{circle}{\kern .01em ~x~\kern .01em} \overset{\small m, b}{\longrightarrow} \enclose{circle}{\kern.01em y~\kern .01em}\) to represent the same concept (in this case, a line with slope \(m\) and intercept \(b\)!).
Reference Texts
In contrast to the books in the previous section, these books are not “the” textbooks for the class! These are here instead as reference books, to keep on hand (a) for when the lectures or the above books are unclear on some topic, and/or (b) for deeper dives into certain topics (where the latter may become a more relevant mission for you as we move towards the final project 😉)
Pearl (2000), Causality: Models, Reasoning, and Inference
As mentioned during Week 2, this is the book containing the fully-developed “unified theory” of causality, starting from a set of axioms and deriving the possibilities of causal inference as formal theorems.
Within “pure” mathematics, if you kept digging into the foundations of things like calculus or algebra, you would eventually arrive at Alfred North Whitehead and Bertrand Russell’s Principia Mathematica—Pearl (2000) is that but for causality: the foundational axioms and core theorems are all in here.
Hume (1748), An Enquiry Concerning Human Understanding
This book serves as the philosophical “jumping off point” for causality: you can think of it like, there’s a nice progress-narrative of the human study of causality, that starts with the uncomfortable questions raised in Hume (1748) about the possibility (or impossibility!) of inferring knowledge of causality via inductive reasoning, and culminates in Pearl (2000).
Sperber (1996), Explaining Culture: A Naturalistic Approach
Mentioned in a footnote in Week 1, this is a surprisingly-old book that made waves in certain communities (like, e.g., among people like me who geek about studying culture quantitatively), by essentially proposing a “research program” for the rigorous quantitative/empirical study of culture. In Jeff’s perfect world, this book would spark a progress-narrative in the same way that Hume (1748)
Applied Examples
The Holy Grail (But, Field = Comparative Politics)
Kalyvas (2006), The Logic of Violence in Civil War
This book is essentially… like, when you read stories about people spending decades hand-carving pathways through a mountain using only a hammer and chisel, this is the social science equivalent of that. A painstaking labor-of-love book that checks every single box I can think of in terms of causally-focused computational social science. It’s my model for any research I try to carry out.
History
Acharya, Blackwell, and Sen (2023), “Slavery, Politics, and Causality”, Harvard Kennedy School Working Paper
A rare paper that touches specifically on the veracity of drawing causal inferences from historical data!
Econometric Policy Evaluation
Björkegren, Blumenstock, and Knight (2025), “What Do Policies Value?”, Review of Economic Studies [PDF]
Another “holy grail” paper, but in this case more for the Data Ethics and Policy course than this course. However, it does touch substantially on the issue of associational vs. causal inference, especially in terms of how going towards causality is a “high stakes” endeavor here, since we’re inferring normative ethical values from data (as opposed to descriptive statistics like the magnitude of the causal effect \(X \rightarrow Y\))
Video Resources
To me (as in, given how my brain works), there are certain topics which I’ve spent hours trying to understand via reading, only to realize that there’s some simple diagram or animation out there which “clicks” it in my mind 10000 times more effectively than the reading ever would.
So, to that end, these video resources are just as important as (for some topics far more important than) the resources above!1
Ahrens (2024), “Robust Causal Inference using Double/Debiased Machine Learning: A Guide for Empirical Research”, MZES Methods Bites Seminar Link
Here I literally added it to the resources folder before I realized that it has a discussion of a paper I did during the PhD. So, perhaps leaving it in here is some sort of humble brag, but finding it originally was not! (It’s what came up for me when I searched “Double/Debiased Machine Learning” on YouTube in May 2025 😜)
Miscellaneous / Commonly-Used Books That We’re Not Really Using
King, Keohane, and Verba (1994), Designing Social Inquiry
I don’t really have a substantive critique of this book, or a deep reason for not using it, other than that I find myself getting really bored whenever I try to read it 🙈
References
Footnotes
For similar reasons, audiobooks may provide more effective ways to digest some topics in the course!↩︎