Using Social Media Data for Research: Potentials and Pitfalls (Online Workshop!)
Indira Sen, Dr. Katrin Weller
Date: 09.11 - 10.11.2020 ics-file
Location: Online via Zoom / Course language: English
The activities and interactions of hundreds of millions of people worldwide are recorded as digital traces including social media data from websites like Facebook, Twitter, Instagram, reddit and more. These data offer increasingly comprehensive pictures of both individuals and groups on different platforms, but also allow inferences about broader target populations beyond those platforms. Notwithstanding the many advantages, studying the errors that can occur when digital traces are used to learn about humans and social phenomena is essential.
In this workshop, we propose to combine theory, data and methods to demonstrate both the pitfalls and potentials of digital traces from social media users. On the first day we start with a general introduction into current approaches in social media research and to types of data used for research, including hands-on insights into example datasets. Furthermore, we will demonstrate various automated options available to a potential computational social science research for collecting, preprocessing and analyzing data.
On the second day, we will introduce the audience to a conceptual framework that helps to identify potential sources of errors in digital trace based research, organized by the different phases in a research process such as data collection, data preprocessing and data analysis.
The proposed error framework is based on and inspired by concepts and guidelines of the Total Survey Error Framework (TSE) that is used by survey researchers and practitioners in the social sciences. Both the TSE and our adaption to the specific characteristics of social media data, will help to to diagnose, understand, and avoid errors that may occur in studies that are based on digital traces of humans from the web.
To help understand the utility of the error framework for digital traces, we apply it to diagnose and document errors in existing computational social science studies such as Understanding Political Opinion using Twitter and Using Search Queries for Inferring Health Statistics.
During interactive parts of the workshop, participants will learn to apply the error framework to hypothetical research scenarios (illustratively using social media datasets openly available on the web).
Survey Methodology, Computational Social Science, Digital Traces, Representativeness, Measurement Errors, Data Collection
Participants will hence gain insights on
- typical scenarios for research based on digital trace data from the web including their potentials for social science research
- how to critically reflect on research design in social media or web data based studies
- how to systematically spot and document errors in their studies
Open to people of different disciplines but primarily aimed at those
- who have some prior experience in survey research and want to extend their knowledge on how digital behavioral data might be suitable additional data sources for their research questions
- who have already worked with digital behavioral data from the web and want to learn about additional possibilities to critically reflect on research designs and their limitations.