As we start a new term and get ready for Explore Econ 2020 and other CTaLE events, including Dr Abdullah Al-Bahrani’s seminar on Diversity in the Classroom and his workshop on Financial Literacy for Students, let’s take a look back at one of the highlights of last term, the extremely popular Causal Inference workshop run by #EconTwitter royalty, Scott Cunningham from Baylor University.
Scott visited CTaLE at the end of October 2019 and presented in the Economics Education seminar series talking about designing an undergraduate course on causal inference, something that is still not very common in most undergraduate or masters curricula, despite there being multiple econometrics courses. In addition, he gave a shortened version of his Causal Inference workshop which was well-attended by students, ranging from undergraduates to PhD students, as well as early career researchers. Alongside talking about causal inference, Scott discussed the process of doing empirical research, the “hidden curriculum”, and the different arcs that an academic career might take. One of our finalists, Sam Asher, who works as a research assistant on UCL’s BME Attainment Gap Project and was the winner of Explore Econ 2019, attended the workshop and wrote up his review of the experience:
More than just econometrics, Professor Scott Cunningham’s Causal Inference workshop establishes a mental framework for and a commitment to progressive, open science. The workshop began with what Professor Cunningham describes as the “hidden curriculum” – skills required for success in Economics that aren’t often covered in lecture halls. These workshops are usually delivered over a couple of days, meaning we could only cover a small subset of their usual subject matter. Professor Cunningham’s choice, for an audience of primarily undergraduates and early-career researchers, to focus on these skills over further methods set the tone for the entire four-hour session.
The hidden curriculum is important for two reasons. The first is practical; everyone new to economic research will find value in a guide to good practise in applied work. Yet the second is more important; in a profession dominated by hierarchy and process, a discussion of the skills and professional activities required for success is surprisingly rare. “It’s vital to network actively and self-advocate”, “You never know which email will open doors”, “You’re in the business of persuasion”; these insights are not obvious, especially to first-gen researchers and those from outside the American university system. The workshop is immensely valuable for this provision alone.
From there, we moved onto the workhorses of modern causal inference: the Rubin causal model and differences-in-differences. These are topics with which UCL students will have a degree of familiarity; we cover diff-in-diff as early as the first year Applied Economics course and potential outcomes in the third year optional Microeconometrics module. But the workshop provides more than just the knowledge required to understand these ideas. Professor Cunningham injects all his teaching with narratives from both the dawn of the methods and his own academic life.
Chief amongst those narratives are the accomplishments of John Snow. In the face of cholera outbreaks 19th century London, Snow was sceptical of the “miasma” theory of the disease’s spread that was dominant at the time. Going door-to-door to collect residents’ water provider information, Snow used a differences-in-differences approach to show that households whose provider had recently moved their primary pump to cleaner water were less likely to contract cholera. As Professor Cunningham tells it, this story is about more than an early use of diff-in-diff. Snow represents the values of data-gathering groundwork, statistical analysis, and rigorous causal reasoning.
That story is not alone; more than just maths and methods, the workshop was about the mind set of science, the necessary approach to discern truth from the world we observe. Famously, John Donohue and Steven Levitt’s paper on abortion and crime was found to have a critical coding error invalidating its results. “People remember the error”, Professor Cunningham explained, “But what they don’t remember is the reason the error was found. When Levitt was asked for the code that generated his results, he sent it over immediately. Levitt really is an example of good science”.
This focus on mind set manifests in an abundance of nuggets of wisdom. “You can’t like your results”, Professor Cunningham asserted more than once, “You need to like your data, your design, but you can never like the results”. We had the chance to chat after the workshop and landed on another criterion: “You have to love the question”. Professor Cunningham lives this mantra in the workshop itself, tearing up the results from one of his own papers in front of us. Although a simple diff-in-diff supported his hypothesis, a DDD approach and testing further hypotheses suggested his original findings were spurious.
Perhaps what is most memorable about the workshop is that striking honesty. Professor Cunningham provides a vision of the Economics profession that is able to acknowledge fault, attempts to share tacit knowledge, promotes new scholarship, and puts the pursuit of truth over professional success. After a year of horror stories, the workshop helped reaffirm that Economics is progressing and that there are people within it committed to that progress. Nothing could be more valuable to young students considering further study.