Quasi-experimental methods for social media analysis
COM2 Level 4
Executive Classroom, COM2-04-02
Abstract:
This talk demonstrates how rigorous methods from empirical economics can be applied to establish causality for problems in computer science, even when experimental data or randomized controlled trials are not available. It applies quasi-experimental methods to Twitter data to investigates the impact of a change in social media design features on the linguistic characteristics of political discussions in social media. Findings suggest that doubling the permissible length of a Tweet led to more polite, less informal, more analytical, and overall healthier discussions online. However, the declining trend in the political relevance of these tweets raises concerns about the implications of the changing norms for the quality of political deliberation.
Biodata:
Dr. Kokil Jaidka is a Presidential Postdoctoral fellow at Nanyang Technological University and is working on profiling users and communities based on their digital footprints. Previously, she was a postdoctoral fellow in Computer Science at the World Well-being Project for two years, and a Senior Computer Scientist with Adobe Research for about three years. Kokil obtained her PhD in Information Studies from Nanyang Technological University in 2014. She has published her work in the proceedings of premier computational linguistics and computational social science conferences and peer-reviewed communication research journals, and is co-inventor on nine patents.