Interview with Felix Lennert

Institute for Political Science, Professorship for Political Science Methods.

Quantitative social science methods with R – especially “data wrangling”, which is an elementary part of the research process, but often ignored.

In the “Big Data Analysis with R” course, we do not analyze “Big Data”, but we talk about the extent to which its potential can be used by social scientists. In addition, the students get a brief introduction to Text Mining as an analysis method. Reference is also made to other possibilities (which then go in the direction of Machine Learning) – but that would go beyond the scope of the course.

Since all of my students come from the social sciences, prior knowledge is such a thing – some already have a little prior knowledge of R and statistics, others cannot even interpret a regression table. My approach here is an applied one: I try to put the students on their own feet as quickly as possible so that they can do everything in R by themselves. The problems they encounter here – and hopefully solve – are part of the learning process and help them to think through the content by themselves. I also evaluate the previous knowledge of my students before the first session in order to know where to pick them up.

This independent approach is not necessarily successful for everyone – last semester some students gave up early, which was probably caused by the conditions (COVID-19, online teaching, etc.) – but those who accomplished the course learned a lot. Among these students were also some without any prior knowledge.

This semester I am fortunated to have a fairly balanced number of students with and without previous knowledge in R in both courses. I divided my students into mixed learning groups (with at least one student with previous knowledge per group). So the tasks are solved in dialogue – and we prevent also a certain isolation this winter. Many students do not even know each other or are new to the city.

Introduction to R, data manipulation with tidyverse, data visualization with ggplot2 and communication of results with RMarkdown in both courses. In one course: “functional programming” with R, basic Text Mining and basic network analysis. In the other course: dealing with survey data, OLS regression, fixed effects regression and difference in difference models.

It’s the future.

I am working on a project called “Mining for Meaning” at Linköping University in Sweden. Here I work with large text corpora and am currently primarily responsible for the acquisition (“scraping”) of the data…

…Furthermore, I am also working on a procedure for identifying relevant postings with the help of a computer (we work with data from an online forum). Here I rely on supervised machine learning.

I attended a summer school as part of my bachelor’s degree. The rest I learned through self-study.

Not since I found a good access to it with R – that is exactly the access I try to convey to my students.

As a social scientist, I work with relatively “simple” theories about human behavior. It is fascinating how these “fancy” methods can be used to properly validate phenomenas about which you already had a fixed idea.

No.

The computational turn should be carried out in every (empirical) discipline, including, for example, history. All departments can benefit from it in some way.

By completing this questionnaire?

Not now.



Mr. Lennert, thank you very much for taking the time to answer these questions. We wish you a nice day!