Ethan Bueno de Mesquita - Thinking Clearly with Data: A Guide to Quantitative Reasoning and Analysis

Thinking Clearly with Data

Thinking Clearly with Data

A Guide to Quantitative Reasoning and Analysis

ETHAN BUENO DE MESQUITA ANTHONY FOWLER

PRINCETON UNIVERSITY PRESS

Princeton and Oxford

Copyright 2021 by Princeton University Press

Princeton University Press is committed to the protection of copyright and the intellectual property our authors entrust to us. Copyright promotes the progress and integrity of knowledge. Thank you for supporting free speech and the global exchange of ideas by purchasing an authorized edition of this book. If you wish to reproduce or distribute any part of it in any form, please obtain permission.

Requests for permission to reproduce material from this work should be sent to

Published by Princeton University Press

41 William Street, Princeton, New Jersey 08540

6 Oxford Street, Woodstock, Oxfordshire OX20 1TR

press.princeton.edu

All Rights Reserved

Library of Congress Cataloging-in-Publication Data

Names: Bueno de Mesquita, Ethan, 1974 author. | Fowler, Anthony, author.

Title: Thinking clearly with data: a guide to quantitative reasoning and analysis / Ethan Bueno de Mesquita and Anthony Fowler.

Description: 1st. edition. | Princeton: Princeton University Press, 2021. | Includes index.

Identifiers: LCCN 2021011897 (print) | LCCN 2021011898 (ebook) | ISBN 9780691214368 (hardback) | ISBN 9780691214351 (paperback) | ISBN 9780691215013 (epub)

Subjects: LCSH: SociologyStatistical methods. | SociologyMethodology.

Classification: LCC HM535.B84 2021 (print) | LCC HM535 (ebook) | DDC 302.01/5195dc23

LC record available at https://lccn.loc.gov/2021011897

LC ebook record available at https://lccn.loc.gov/2021011898

Version 1.0

British Library Cataloging-in-Publication Data is available

Editorial: Bridget Flannery-McCoy, Alena Chekanov

Jacket/Cover Design: Wanda Espaa

Production: Erin Suydam

Publicity: Kate Hensley

Copyeditor: Elizabeth J. Asborno

Cover image: Igor Kisselev / Alamy Stock Vector

Ethan: for Abe and Hannah

Anthony: for Gloria

1. xvii

1. xvii
2. xviii
3. xix
4. xx

The world has changed in transformative ways. Data and evidence are ubiquitous. Quantitative information suffuses our talk of everything from policy to health care to job searches to politics to sports to education to dating to national security.

As a result, statistics and quantitative reasoning must no longer be the purview of only those who have a knack for mathematics or are headed for technical careers. Acquiring competence in foundational quantitative reasoning is now a fundamental responsibility of every educated human being and citizen. And this necessitates new ways of teaching and learning.

It was with that goal in mind that we decided to write Thinking Clearly with Data. But we didnt start with the book. Much of the material and ideas that ultimately found their way into the coming chapters were first developed for courses aimed at providing to students with little technical background the tools needed to be serious, thoughtful, and skeptical consumers of quantitative information. These courses include traditional university offerings, like introductions to quantitative reasoning taught to both undergraduate and graduate students at the University of Chicago. But they also include executive education courses offered to policy makers, military officers, national security experts, intelligence professionals, and journalists.

We learned a lot of lessons along the way that inform the choices we made in writing and organizing this book. Perhaps the most important was to create a shared language.

We knew we didnt want to teach a traditional statistics course. Such courses, in our view, are often too technical for many students and dont get to the most important and interesting issues, the ones that really matter for using quantitative information to make our lives and the world better. So, it was tempting to jump as quickly as possible to the exciting topics, like why correlation doesnt imply causation. But that would have been a mistake. A person cant understand why correlation doesnt imply causation until they understand what correlation and causation are.

For that reason, of this book is all about establishing a shared language. We define, conceptually and technically (but still accessibly), what we mean when we talk about correlation and causationnot in the sense of how to calculate a correlation coefficient or how to write down a causal effect in potential outcomes notation (though both will be covered), but in the sense of the questions, What do the words, properly understood and digested, mean in plain English? Whats hard about correlation and causation? Why are they usefully separated? What are these two kinds of things, correlations and causal relationships, good for?

But what about the problem of motivation? If you dont put the good stuff up front, how do you keep people engaged? Well, first, who says a conceptual understanding of what causality does and does not mean isnt the good stuff? It is great stuff. But, more to the point, our approach is this: if you want people to be engaged, make the material engaging. To us, this means several things.

The first is to tell stories. Throughout, you will find every conceptual discussion augmented by at least one extended, genuine, real-world example. Some of those examples will be about scientific studies. Many will be about personal experiences of ours where thinking clearly about quantitative evidence made a difference in the decisions we made. Others will involve reflections on the use of data and evidence in news, sports, policy, health care, and culture. This stuff really matters for how lives are lived and decisions are made in every realm of human endeavor. We want to keep that fact in the foreground. That is also why, despite the fact that this is a book by two political scientists, many of the examples are not drawn from politics.

The second way to engage readers is to emphasize ideas first and technicalities second. We love technicalities. But technicality can often be the enemy of understanding. When things get technical, lots of people stop thinking and start memorizing. We fervently wish to avoid that. So we always talk about the ideas and why they matter first. We treat things graphically whenever we can. And we do as little math as possible. But as little as possible isnt zero, for at least two reasons.

Familiarity with some technical matters is part of being a clear thinker. You cant understand mean reversion if you dont know what a mean or noise is. You cant understand publication bias and the replication crisis if you dont know what statistical significance means or what a p-value is. And it is hard to understand the problem of confounding or the answers offered by different research designs without being able to interpret a regression.

Moreover, sometimes being clear and precise requires a bit of math. We spend lots of time talking conceptually about counterfactuals and causality. But counterfactual talk can get a bit mystical. There is an extra degree of clarity that comes from writing down some potential outcomes and a proper definition of an effect that we think is indispensable. So we do not dispense with it. But, always, our emphasis is on clear thinking.

A third lesson for engaging writing is that it isnt enough for each chapter or lesson to tell a story. The whole course (or book) must do so. For us, the story is that making good decisions and doing good in our data-driven age requires clear thinking from each and every one of us. We cant just leave it to the experts, for many experts were never taught to think clearly about quantitative information. So we have to do it for ourselves or we will be frequently misled and may well make terrible mistakes.