ICPSR (Inter-university Consortium for Political and Social Research), which owns the world’s largest archive of research data on social sciences, holds an annual summer program in Quantitative Methods of Social Research at the University of Michigan, where it is headquartered. The ICPSR Summer Program, which has continued since the 1960s, was held online this year as well as last year, with four weeks of intensive lectures on more than 40 courses.
The Center for Positive/Empirical Analysis of Political Economy subsidizes tuition fees every year for graduate students participating in this summer program. This year, with our support, a total of three graduate students from the Graduate School of Political Science and the Graduate School of Economics worked on this four-week program.
Affiliation/Grade: Graduate School of Political Science, D1
Research Interests: Modern politics
Course(s) Attended: Data Science and Text Analysis, Time Series Analysis II: Advanced Topics, Introduction to the LaTeX Text Processing System, Introduction to Python, Categorical Data Analysis, Introduction to the R Statistical Computing Environment
The course “Data Science and Text Analysis” aimed to develop theoretical background, practical experience, and technical competence to pursue cutting-edge social science research using textual data. This course provides key technical aspects for conducting research using textual data, with a particular focus on technologies drawn from new areas of data science, from data collection and preprocessing to explanation and inference analysis.
At the end of the course, I acquired the basic skills needed to perform data collection and management tasks that you may encounter during a research project intended to be published as an article in an academic journal. I also acquired the skills and experience of collecting and preprocessing digital text data and applying many standard text analysis techniques to the data.
I believe that the experience gained in the course can be used for text analysis and content analysis of textbooks. In particular, I think that it is possible to directly utilize the creation of data frames using R and the word cloud that connects related words for research using my own text analysis.
People who are not native English speakers may hesitate to take the course, but the teacher does not ask students questions individually, and the screen is allowed to be turned off, so it seems that the mental burden is quite small. In addition, one of the functions of the ZOOM used at the University of Michigan is called “transcript”, which allows you to instantly display the words spoken by the teacher. If there was a word I didn’t understand, I could translate it on the spot, which was very convenient. I learned about this feature about two weeks into the course, so I wish I knew about it sooner. The best thing is that the lessons are online and the recordings are distributed throughout the lecture, so I was able to take the lessons anytime, anywhere. Even after the course is over, you can watch the recording for about two weeks.
Affiliation/Grade: Graduate School of Political Science, M1
Research Interests: Comparative politics
Course(s) Attended: Race, Ethnicity, and Quantitative Methodology I
This course provided students with an overview of the major theories and empirical approaches to the study of intergroup attitudes. It has also devoted considerable amount of time in methodologies employed in the study of intergroup attitudes. Since most of the debates on race and ethnicity revolve around measurement, the content focused on different methods in scaling and dimensional analyses, and their applications in the corresponding literature. Each week the lecturers began with a theoretical discussion, then continued with methodological lectures and occasionally involved some guest speakers on relevant topics.
Via joining this course, it has enhanced my understanding on the theory and measurement of ethnicity and race. By asking questions to lecturers directly and getting assignment feedbacks from TA, I have also learnt more about my interested subjects and attained some improvement in my current research work on survey design and IRB preparation. They are friendly, respectful to students from different backgrounds. Although it is hard to connect with other students online on zoom lectures, it is still a nice experience to be in the same class with them who are from different racial backgrounds of Black, Latino, Asian, White and indigenous groups.
Advice for future students: the course focused more on US-context theoretically and empirically with a generalized framework on race and ethnicity and discussion of literature mainly in the US, so if you are not working on US-related topics, this course may not be the best choice for you. Instructors welcome questions from students, so bring your own research problems to the course would be the best way to learn from it. They also put the lecture videos available online 14 days after the lecture, so you may also revise the content again if you want to.
Affiliation/Grade: Graduate School of Political Science, M1
Research Interests: International trade
Course(s) Attended: Machine Learning: Applications in Social Science Research
A growing number of social scientists are taking advantage of machine learning methods to uncover hidden structure in their data, improve model predictive power, and gain a better understanding of complex relationships between variables. This class covers the mechanics underlying machine learning methods and discusses how these techniques can be leveraged by social scientists to gain new insight from their data. Specifically, it covers both supervised and unsupervised methods: decision trees, random forests, boosting, support vector machines, neural networks, deep and adversarial learning, ensemble learning, principal components analysis, factor analysis, and manifold learning/ multidimensional scaling. In this course, we also discuss best practices in fitting and interpreting these models, including cross-validation techniques, bootstrapping, and presenting output. The most interesting part is that the workshop demonstrates how models can be estimated in R.
When I prepared for the CFA examination last year, I learned some basic knowledge about machine learning, but it is not enough for me to know the theory only, I should know how to use it. And this class offers me a great chance of deepening my understanding of machine learning method in the context of code or R. As the development of AI, in the future, maybe some jobs would be replaced by robots. On one hand, people should master some skills that is irreplaceable. On the other hand, people should learn how to create and use AI. For this purpose, learning programming language and machine learning will be a small step. In the United States, there are four famous universities in the field of computer science: MIT, Sandford, UCB and Carnegie Mellon University. After graduating from WASEDA, I plan to work in a financial company for three to five years and try to obtain an offer of computer science in one of the four universities in America. I think this second master degree will help me switch my direction and walk further in the field of finance. I really appreciate our school, offering us such a valuable opportunity of participating in the summer school in U-Michigan and learning what I need for my future plan.
And now I am writing my master thesis on international trade, more specifically, the paper is about the trade war and the best tariff. It is a theoretical paper, but I need to use real world data to verify my model, and it is obvious that this course offers me a new sight of uncovering the relationship of data.
If someone is interested in Machine Learning or R, I strongly recommend this course.