Methods Bites

Blog of the MZES Social Science Data Lab

Computationally Analyzing Politicians’ Body Language Using Pose Estimation

2025-01-10 25 min read tutorials [Oliver Rittmann]
Politicians can increase the appeal of their speeches through nonverbal cues such as gestures and vocal emphasis. Understanding the factors that make political speech appealing is central to political science research, yet studying nonverbal cues during political speech is difficult due to their audiovisual nature. Pose estimation models—a class of computer vision models that locate and trace human body key points, such as hands, ellbows, and shoulders throughout videos—offer a valuable opportunity to computationally assess politicians’ body language in video recordings. Continue reading

Using TikTok ads for survey recruitment: a step-by-step approach

TikTok’s rapid growth and diverse user base present social science researchers with a unique opportunity to study a large and varied population, gaining valuable insights into their attitudes and behaviors. Unlike platforms such as Facebook and Instagram, TikTok’s potential in survey recruitment has been relatively underexplored. The platform’s cost-effective reach and detailed targeting parameters make it particularly appealing for reaching traditionally hard-to-reach or rare populations. Furthermore, with its video-centric format and predominantly young user base, TikTok provides a means to engage and attract respondents from younger generations to participate in online surveys. In this Methods Bites Tutorial, Zaza Zindel and Simon Lütkewitte (Bielefeld University) provide a step-by-step guide on how to use TikTok ads for survey recruitment. Continue reading

A Hands-On Introduction to Artificial Neural Networks

2023-07-18 26 min read tutorials John 'Jack' Collins

Neural networks are powerful machine learning algorithms that form the basis of many important technologies, including generative AI and computer vision. However, they are not as straight-forward to implement as many other machine learning techniques, like random forest or logistic regression. If you are a researcher interested in applying neural networks, this Methods Bites Tutorial by John ‘Jack’ Collins demonstrates how to get started with Artificial Neural Networks (ANN) and helps you easily prototype a neural network for your own use-case. Continue reading

BERT and Explainable AI

2023-03-28 43 min read tutorials [Andreas Küpfer Cosima Meyer]

Natural language processing (NLP) is a fascinating field. Popular NLP techniques for understanding (written) human language include next-sentence predictions, translations, text classifications, or sentiment analysis. Such techniques already permeate our everyday lives: What would the world be without services such as Google Translate, DeepL, or the recently released ChatGPT? While common bag-of-words approaches can often be a valuable approach for NLP, Google’s release of BERT in 2018 revolutionized the possibilities in NLP. This Methods Bites Tutorial introduces the logic of large language models (LLM) with a special emphasis on BERT. It provides an applied use case from the social sciences, walks readers through explainable artificial intelligence (AI), and explains how we can leverage explainable AI to explain predictions of our models. Continue reading

Collection, Management, and Analysis of Twitter Data

2022-06-02 18 min read tutorials Andreas Küpfer

As a highly relevant platform for political and social online interactions, researchers increasingly analyze Twitter data. As of 01/2021, Twitter renewed its API, which now includes access to the full history of tweets for academic usage. In this Methods Bites Tutorial, Andreas Küpfer (Technical University of Darmstadt & MZES) presents a walkthrough of the collection, management, and analysis of Twitter data. Continue reading

Survey data collection from start to finish

2022-04-11 22 min read instructionals Joshua Hellyer

Surveys have long been a staple of social science research on individuals’ attitudes and behaviors. In recent years, however, we have witnessed a strong shift from secondary analyses of large general social surveys toward smaller, more targeted primary data collections. This development has been accompanied by the increasing availability of affordable and easy-to-implement surveys using online access panels. While the entry barriers to original survey-based research are now likely lower than ever before, it still comes with notable methodological, administrative, and logistic challenges. To help aspiring survey researchers navigate this process, this Methods Bites Instructional by Joshua Hellyer (MZES, University of Mannheim) provides a comprehensive guide to survey data collection with online access panels. Continue reading

Telling Stories with Data: Insights into Data Journalism

2021-11-12 10 min read instructionals [Yannik Buhl]

Telling stories with data is one of the most important things to do for almost everyone working with data analyses. Why? Because its goal is reaching the audience one wants to reach. If one succeeds therein – be it, for example, average news consumers or academics in a specific field –, the underlying data analysis will more probably have a lasting impact. A good story drags the audience into your analysis. In this Methods Bites Tutorial, Yannik Buhl offers a recap of his workshop “Telling Stories with Data: Insights into Data Journalism” in the MZES Social Science Data Lab during Spring 2021. It focuses on the important steps to tell a thorough story based on data analyses – and how scientists and data journalists can learn from each other. Continue reading

Older posts