Does ‘X’ Really Cause ‘Y’?

Summer 2020


In an age when many stories in the reported news exaggerate claims and use bold language that doesn’t match actual data, a course offered last January over Intersession couldn’t have been timelier. Its aim: to help students detect biases and flaws in what they read.

During the first week of Should Susan Smoke? An Introduction to Causal Inference, the nine students in the class examined a New England Journal of Medicine study that claimed a correlation between chocolate consumption per capita and the number of Nobel laureates in a selection of countries. The students were shocked that such a study could be published.

“I read it as a joke!” one student chortled.

“It definitely wasn’t completely serious, was it?” another asked as she shook her head in disbelief.

While an extreme example, the course instructors say kooky correlations like the one highlighted in this study can appear anywhere, and students need to be vigilant in scrutinizing the world around them.

“We want students to learn how causal inference can help them reason through important issues, such as the approval of new drugs and treatments, climate change, education policy, and the criminal justice system,” says Razieh Nabi, a graduate student in the Whiting School’s Department of Computer Science and one of the course’s instructors.

Nabi and course co-instructor Rohit Bhattacharya, also a graduate student in computer science, say their ultimate goal in teaching the class was to help their students become more scientifically literate and informed citizens.

“Papers come out all the time that say ‘x’ is associated with ‘y,’ like, drinking wine and getting cancer,” Rohit says. “We’re teaching students to pause and ask, ‘If I do x, will it really cause y?’”

Some of the other case studies up for discussion in the course included whether smoking causes lung cancer and whether estrogen causes endometrial cancer. Students also examined name-swapping experiments in hiring discrimination cases to more clearly conceptualize what fairness means in the legal system and how antidiscrimination laws use the language of causal inference.

While there is a more advanced course in causal inference at Johns Hopkins, Nabi and Bhattacharya’s Intersession course was geared toward undergraduates because they believe it’s essential for students to have these reasoning tools as early as possible.

Eventually, they hope their course will become offered regularly during Intersession and perhaps someday even adapted for high school classrooms.