By Eshima Shusei
About the Program
- Program: the University of Michigan Inter-university Consortium for Political and Social Research
- Course: Designing, Conducting, and Analyzing Field Experiments
- Instructor: Professor Donald P. Green
I attended ICPSR course “Field Experiments” taught by Professor Donald Green. In the lectures, we started with the basic of causal inference framework named potential outcomes framework, which is widely used not only in Political Science but also in other science field such as epidemiology. In potential outcomes framework, at least two potential statuses are supposed for each unit of analysis: one when get treated, and one when not. Naturally, we cannot observe both statuses for each unit at the same time, so we consider the average treatment eﬀect by subtracting the average of outcome in untreated from the average in treated under the condition that the treatment is randomly assigned.
Though it may sound simple, randomly assigning treatment in social science setting is hard. The second part of the course was spent on how we can design a proper research so that randomization scheme is evident and justifiable. This part was based on his recent research in Uganda, and using the actual case, he introduced several ways to randomize, such as a block randomization and a cluster randomization, along with advantages and disadvantages. The third day was spent on discussing participants’ future research proposal. There were only seven students, but we had a variety of backgrounds and interests from a Ph.D. student studying online media to an NGO staﬀ evaluating their program. My plan has a funding problem if I want to assign the treatment randomly by myself, so he advised me how to utilize existing surveys.
The summer course itself was great, and it was a great opportunity for me to talk with other participants who are also interested in experimental methods but studying other topics. For example, one of them was interested in analyzing how discussions go on in online platforms. He advised me how to handle a large size of data in the analysis, which I was looking for these couple of weeks. Another student studies political behavior in an election campaign. He taught me practical problems he faced during his project. Since it is hard to find students who are interested in empirical studies in general in Japan, this course was a great opportunity for networking as well.
I would like to express the greatest gratitude to the generous scholarship of FUTI. Without your support, I could not have a chance to attend this great course and meet wonderful colleagues.