Experiments without experimenting
Guido Imbens uses data science to search for cause and effect in the wild.
Story tags:
Saturdays were laundry day.
It was Guido Imbens’ first year teaching at Harvard, and with fellow economist Joshua Angrist, a weekly tradition had taken hold at their local laundromat. While the machines rumbled, the two men would discuss current research and generally bat around questions they were chewing on. Among them: How might cause and effect be revealed using natural experiments—observational studies, that is, that emerge in the wild rather than from the researchers themselves?
As their friendship deepened, their ideas did, too. In 1994, they published a groundbreaking paper explaining how to draw causal inferences from observational data. Then, in 2021, the two received calls that would change their lives. Those Saturday conversations in the laundromat had spun, decades later, into a shared Nobel Prize in Economics. (Imbens’ donation to the Nobel Prize Museum? Laundry detergent.)
Today, Imbens is the Applied Econometrics Professor of Economics at Stanford Graduate School of Business. In essence, he wades into real-life phenomena—social, financial, economic—and susses out cause and effect. His ability to find causality in deeply complex scenarios has made him a go-to resource among policymakers; his data and statistical work provides insight where designed experiments aren’t feasible. It would be unethical to measure the effect of education on labor market outcomes or earnings, for instance, using a social experiment that deprived some subjects of a good education.
“It’s very hard to figure these things out because you typically can’t do experiments.So we need to work with observational data, which makes it very challenging to separate correlation from causation,” he says. “What I do, in general, is come up with methods for disentangling the two.”Guido Imbens
By using existing data to examine economic outcomes, we are in effect conducting “experiments without experimenting,” Imbens adds, tossing in a quote from the poet Virgil for good measure: “Happy is he who knows the causes of things.” But Imbens does not seek knowledge for knowledge’s sake. His work helps shape pivotal societal policies and regulations, and has looked at problems ranging from the impact of minimum wage on unemployment to the effectiveness of a new drug on a patient.
“For economists, we’re motivated by trying to make the world a better place and trying to figure out how we can ensure that economic policy helps all of society,” Imbens says. “At some point the particular question just grabs me. And then I have a hard time letting go. I’m fascinated by the problem and that’s really what motivates me to pursue it.”
“For economists, we’re motivated by trying to make the world a better place and trying to figure out how we can ensure that economic policy helps all of society. At some point the particular question just grabs me. And then I have a hard time letting go. I’m fascinated by the problem and that’s really what motivates me to pursue it.”Guido Imbens
That’s how he came to start the Stanford Causal Science Center, part of Stanford Data Science, a university-wide initiative that is advancing data-intensive methods and research across disparate fields, from health to sustainability. His focus: using mathematical and statistical techniques to elucidate the causes of social, financial, business, and economic phenomena through observation.
“For me, it’s always been very important to talk to people in other areas and see what they’re working on, to listen to them and figure out where they may actually need new methodologies,” Imbens says.
Focus on the future
The bad news: The world is complicated. More so every day, it can seem, as environmental crises thread through economic crises, which thread through health crises, and so on.
The good: Data science is here for it. As interwoven and impacted as these crises are, our ability to understand them is exploding. Imbens regards his work at Stanford as preparing the next generation for the biggest challenges in human health, economics, social justice, and sustainability—realms that require clarity as much as they require action.
“I see a great deal of optimism among the students, despite the fact that these are clearly incredible challenges. Especially climate change, where still we haven’t made a lot of progress. Stanford is putting huge resources into that, and you see the students being very excited about the new school and the possibilities of interdisciplinary work, helping them solve these problems,” he says.