Lecturer

• Lecturer at the Department of Sociology with a Focus on Methods of Quantitative Empirical Social Research
• Research interests: Migration & integration, inter-group conflict, attitudes, mass media, political communication
• Methods: Classical quantitative methods for social research combined with computational methods and natural language processing
• More info on my homepage

Office Hours

• After appointment
• Office: 3.G152 (PEG)
• Contact me at cc@soz.uni-frankfurt.de
• Do not hesitate to write me if you have questions, comments, doubts, criticism etc.

General information

• Thursdays, 14:15 in room PEG 2.G 111 / 116
• Material available on GitHub
• Join this course’s mailing list for communication
• Course on QIS
• Slides in English, Kurs auf Deutsch 🙂

What is CSS?

• “The intersection of social science and data science is sometimes called computational social science.” (Salganik 2018: xviii)
• Answer social science questions with computer science methods
• New, “big” data sources
• New methods, much more computational power
• In particular, text mining and natural language processing (NLP) have received increasing attention in the social sciences

Definining big

• 3 Vs: Volume, Variety, Velocity
• A lot of data…
• In a variety of formats…
• Being created constantly

The boom of CSS and NLP

• Books Ngram Viewer
• “Google Books Ngram Viewer, it displays a graph showing how those phrases have occurred in a corpus of books” google.com

Goal of seminar

• The goal is that students learn how to use computational tools to analyze big and small data
• The training will focus on R, but we might work with Python from time to time
• This implies continuous work throughout the semester

You should have…

• Interest in quantitative social research and programming
• Good working knowledge of descriptive and inductive statistics (regression etc.)
• Some knowledge of R or another statistics software / language
• I will introduce both in this seminar, but the focus will be on computational methods

Your interest

Average motivation is: 9.3157895/10

Your knowledge: Familiarity with CSS

Prior knowledge CSS

Other prior knowledge

• Percentage who ran a logistic regression before: 0.8421053
• Percentage who used machine learning before: 0.2631579
• Average experience with code-based data analysis: 3/5
• Percentage who usually work with R: 0.5263158
• Percentage who usually work with Python: 0.2631579
• Percentage who never use any statistical software: NA

Your expectations (selection)

[1] "Ich wünsche mir, dass die Anforderungen an die Student*innen von Anfang klar kommuniziert werden bezüglich des Levels, das wir uns bei R (möglicherweise noch vor dem Kurs) selbst aneignen müssen und wie die Teilnahmenachweise und Leistungsnachweise erbracht werden."

Your expectations (selection)

[1] "Textanalysen von Internetquellen bzw. Textdokumenten (politische Positionspapiere), sofern  dies in den betreffenden Methoden enthalten ist. Jedenfalls verstehe ich dies so."

[1] "My personal main area of interest lies with text mining / natural language processing methods. I am especially interested of running analysis with something like sentiment analyis as I learned how to do it in my BA, but never got around to do it in earnest."

[1] "Ich habe schon einiges an linearen und log regressionen ausgeführt und würde meinen Werkzeugkasten gerne erweitern. Am liebsten würde ich weiterhin in R arbeiten."

[1] "Ich habe Kenntnisse in Statistik und bereits Erfahrungen mit Python gemacht, allerdings weiß ich noch nicht genau mit welchen Methoden beides kombieniert werden kann. Daher würde mich insbesondere der methodisch-soziologische Bezugspunkt interessieren."

Your expectations (selection)

[1] "I am wondering how to get access to \"big\" data. Social science data is often very limited so I am still wondering how to apply machine learning for social phenomena"

What this course will offer

• A guide to the scraping, preparation, and analysis of different types of digital trace data (focus on text-as-data)
• The means to conduct your own computational research project
• Hands-on application of methods in tutorials

What this course will not offer

• In-depth understanding of mathematical foundation of methods
• Advanced programming
• Course is not suited as a general introduction to quantitative social science methods

Literature

• See GitHub syllabus for literature on individual sessions
• Van Atteveldt, W., Trilling, D., & Calderon, C. A. (2022). Computational analysis of communication.
• Salganik, M. J. (2019). Bit by bit: Social research in the digital age. Princeton University Press.

Term paper

• Computational analysis of digital data (in the broadest sense) on a research question of your interest
• You can choose from any topic (… for which you find suitable data)
• The focus of the paper should be the analysis

Term paper

• ~15 to 20 pages, incl. tables, graphs, references etc.
• Also add your code so that I can understand and reproduce what you’ve done
• DEADLINE IS 01 APRIL 2025!
• Hand in as PDF and code via e-mail to cc@soz.uni-frankfurt.de

Syllabus overview

Syllabus overview

Syllabus overview

Syllabus overview

Syllabus overview

Syllabus overview

More information about each session on GitHub

Data collection

• Download data sets from websites, online archives, repositories, …
• Application Programming Interfaces (APIs)
• Web scraping
• Collect your own data

Data analysis

• Sentiment analysis
• Text summary
• Machine translation
• Classification
• Regression etc.

Text analysis

Boumans & Trilling(2016)

Example 1

Political migration discourses on social media: a comparative perspective on visibility and sentiment across political Facebook accounts in Europe

Example 2

Real-World Developments Predict Immigration News in Right-Wing Media: Evidence from Germany

Example 3

The rise of Jihadist propaganda on social networks

Example 4

Online hatred of women in the Incels.me forum Linguistic analysis and automatic detection

R

• You will need R for most tutorials and the term paper (Python use is encouraged as well)
• To work with R, install on your computers
• R at cloud.r-project.org
• RStudio at rstudio.com

Python

• To work with Python, install on your computers
• Python at python.org
• Anaconda at anaconda.com (email required)
• Important: Python does not like white spaces in your file path (i.e. “C:\Users\Christian Czymara\”)

GitHub

• Material will be uploaded on GitHub
• github.com/czymara/CSS_WS24
• You can download files without having an account
• For advanced users: Feel free to make an account and download GitHub Desktop to synchronize files every week

ChatGPT

• Not strictly necessary, but can greatly help you code and solve coding related problems
• Requires a (free) OpenAI account
• In particular, check out Code Nerd (requires ChatGPT Plus subscription)

R

• Why “R”?
• “R is an implementation of the S programming language” (Wikipedia)
• R is a programming language for data analysis
• Rstudio is a so called Integrated Development Environment (IDE), making writing and running R code a lot easier
• Overview of stored objects
• Projects containing multiple files
• Git connection
• Etc.

R benefits

• Free and open source
• Large and very helpful community
• Plethora of user-written packages on basically everything
• Very powerful tools for data manipulation and data visualization
• In addition to analyzing data, you can write programs, websites, books, and much more with R (and R Markdown)
• … and integrate with other languages

Google Colab

• To understand the basics of R, we will work with this Google Colab

Google Colab

• To understand the basics of Python, we will work with another Google Colab

Tasks for next week

• Install R and Rstudio
• Familiarize with R
• Bring one of your term papers or your BA thesis in DOCX format on a USB stick
• Optional: Install Pyhton and Anaconda

Literature

• The exercise is inspired by this great intro from Fabio Votta
• Wickham & Grolemund (2017). R for Data Science. O’Reilly