Unveiling influencer-driven PII disclosures in social media discourse
Eidan J. Rosado, Ling Wang, Laurie Dringus, Junping SunPurpose
This study examined how influencer power and tier relate to social media engagement and personal information sharing, noting fluctuations of both over time.
Design/methodology/approach
A combination of content analysis, change point detection methods (CUSUM Drift, Change Point Detection) and time series modeling (ARIMA) was used to analyze social media conversations and identify temporal trends in engagement and personal information disclosures.
Findings
The study demonstrates a strong correlation between influencer reach, increased engagement and disclosures, with a declining trend in both engagements (79.78% of threads) and disclosures (66.12% of threads) across 183 conversational threads. An observed 29.41% of posts contained personally identifiable information (PII), with conservative sensitivity analysis suggesting an adjusted prevalence of 15.00% to 20.27% after accounting for automated detection error.
Research limitations/implications
The cross-platform comparison revealed architectural differences between Twitter/X's follower hierarchy and Reddit's community voting structures that influence influencer effects on PII disclosure. Rethinking influencer identification on community-based platforms requires tailored models that consider community norms instead of just follower counts. Manual validation of automated PII detection was infeasible due to data access constraints.
Originality/value
This project provides exploratory insights into how platform architecture may moderate influencer dynamics, with implications for privacy-conscious platform design and future comparative studies.