Abstract
This replication study presents a methodological replication of the work by Chau et al. (2020) that was originally published in MIS Quarterly. The original study adopted the design science paradigm to develop a novel lexicon through collaboration with clinical psychologists to analyze emotional distress in social media content and combined machine learning (ML) and rule-based classifiers to enhance detection accuracy in Chinese blogs. Our methodological replication seeks to implement and apply these methodologies to English-language posts from Facebook. Our key findings reveal that our ML classifier with the Linguistic Inquiry and Word Count (LIWC) lexicon achieved the highest recall of 97.59% and an F1-score of 97.54%. However, these results differ from the original work, where an ensemble of ML and rule-based classifiers performed best. We elaborate on possible reasons for these differences, including language, culture, content length disparities between the datasets, different social media platforms, and intricacies of translating lexicons across languages.
DOI
10.17705/1CAIS.05749
Recommended Citation
Yuan, A., Gao, Y., Garcia Colato, E., & Samtani, S. (2025). Identifying Emotional Distress on Social Media: A Replication Study. Communications of the Association for Information Systems, 57, 1134-1146. https://doi.org/10.17705/1CAIS.05749
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
