INSTITUTE FOR POLICY RESEARCH

Faculty Spotlight: Charles F. Manski

‘Unlearning and discovery’ to improve public policy

IPR econometrician Charles F. Manski's work is deeply rooted in methodology, but inexorably tied to policy and motivated by real-world problems.

A body of nearly 150 peer-reviewed articles and eight books attests to the wide-ranging research of IPR econometrician Charles F. Manski. Deeply rooted in methodology but inexorably tied to policy and motivated by real-world problems, his publications cover topics from analysis of choice behavior and treatment effects to the death penalty and who attends college.

His most influential econometric work, ranging from development of the maximum score estimator to study of the reflection problem to analysis of partial identification, has either relaunched or upended established literature across many fields beyond economics.

Ask Manski how he came to such field-changing insights, and he describes a process of “unlearning and discovery,” or going “back to basics” in which he questions commonly accepted knowledge or doctrine. It is a process that taught Manski how to “live with ambiguity,” and it has steered his work. His investigations of how one understands and uses data have uncovered tangible, far-reaching effects not just for economics, but society.

Over the course of his career, he has also sought to connect his work to a wider policy orbit.

“I decided that communicating only with mathed-up econometricians or economists was just not going to be enough,” Manski said.

Communicating Uncertainty 

Over the past decade, Manski has turned his attention to how researchers and policy analysts communicate findings to a wider audience. 

He is troubled by the prevalence of “incredible certitude;” that is, making incredibly precise forecasts of policy outcomes—beyond what one should reasonably expect research to deliver. He has emphasized the “law of decreasing credibility,” which cautions that research using unwarranted assumptions to obtain precision yields conclusions that are less credible.

Instead of communicating that a hurricane or recession has a precise 40% chance of occurring, for instance, Manski urges researchers, statisticians, and officials to offer more realistic interval forecasts pulled from a range of plausible models, such as a 30–50% chance that such events could occur. He also advises updating the forecasts as the models evolve and data accumulates.

Convincing economists to be more honest in acknowledging the limits of their data and findings got underway with a 2011 article in The Economic Journal, followed by his 2013 book, Public Policy in an Uncertain World (Harvard University Press). Manski wrote the latter explicitly “in English” for a wider public, limiting the math to one formula in the entire book. He has since tackled uncertainty in government statistics in 2015 and outlined six practices that lead to incredible certitude in 2013.

Manski attributes his efforts to widen outreach beyond academia to his time as director of the Institute for Research on Poverty at the University of Wisconsin–Madison from 1988–91, where he was then on the faculty.

“All of a sudden, I had to write for and speak to all kinds of people,” he recalled, not just academics. There were legislators, government statisticians, program officers, school superintendents—all sorts of people dealing with “real problems in the world.”

Tackling Uncertainty in Medicine 

Following a medical scare in which Manski found himself in an Israeli hospital with a chart marked FUO, or “fever of unknown origin,” in 1985, he began digging into medical decision making. He saw another field that could benefit from his studies of uncertainty and its foundations in econometrics and statistics.

“I think people like myself have something to contribute to evidence-based medical research,” Manski said. “But we better be very careful about it.”

Having no medical background or even high school biology, Manski began building his medical knowledge from zero.

“Medicine is so vast, you can’t be an expert in everything,” he said. He notes the difficulty of medical specialists to know much about fields outside their own, given the ever-growing number of specialties and knowledge. He devoted months to deciphering medical articles, focusing on a few clinical areas, such as breast cancer, hypertension, and organ transplants and now feels confident enough to converse with specialists on these topics.

“You don't want to give me a scalpel or anything, but I am increasingly understanding what the basic issues are,” Manski said. “There's no ‘unlearning’ because I never had [that medical knowledge], but there’s incredible learning.”

The result of his investment in medical knowledge combined with his world-class work in decision making is his forthcoming book, Patient Care Under Uncertainty (Princeton University Press, 2019). In it, Manski takes on several major tenets of medical research and decision making, including hypothesis testing and meta-analysis.

He outlines the ways in which these and other conventional methodological practices are ill suited to designing and interpreting randomized trials, and he explains how they have undermined medical decision making. In their place, he offers innovative applications of statistical decision theory. He sees it as offering a better path to the design and analysis of clinical trials.

Manski hopes the book will allow doctors and health professionals to better understand the ambiguity inherent in medical research so that they can make more informed and credible projections about their patients’ care. 

Widening the Circle of 'Unlearning' 

While Manski spent the first part of his career studying ambiguity for an audience of researchers, he has progressively broadened his scope to policymakers, doctors, central bankers, weather forecasters, and even just to plain folks, encouraging them to embrace ambiguity and uncertainty as well.

“Everybody has to unlearn at times,” Manski said. “It should be totally natural to communicate uncertainty in any kind of policy analysis or anything else.”

Charles F. Manski is Board of Trustees Professor in Economics and an IPR fellow.

Published: August 19, 2019.