Studying Behavioral Addiction by Combining Surveys and Digital Traces: A Case Study of TikTok

TL;DR

This paper investigates behavioral addiction on social media by combining survey data with GDPR-mowered data donations from TikTok usage traces. It adapts the Bergen Facebook Addiction Scale to TikTok, revealing that 27% of participants are highly likely addicted, with young adults disproportionately represented. Through data donations from 107 participants, the study finds that highly likely addicted users spend more time and return to TikTok more often, while content sentiment and in-network engagement do not robustly distinguish addiction levels. A multi-layer perceptron classifier using basic usage features achieves a moderate $F_1$ of at least $0.55$ in identifying highly likely addicted users, suggesting that predicting behavioral addiction from usage alone is challenging and that richer social-context signals are needed. The work highlights methodological feasibility and policy relevance under the Digital Services Act, while noting limitations such as sample bias and the need for cross-platform and qualitative follow-ups.

Abstract

Opaque algorithms disseminate and mediate the content that users consume on online social media platforms. This algorithmic mediation serves users with contents of their liking, on the other hand, it may cause several inadvertent risks to society at scale. While some of these risks, e.g., filter bubbles or dissemination of hateful content, are well studied in the community, behavioral addiction, designated by the Digital Services Act (DSA) as a potential systemic risk, has been understudied. In this work, we aim to study if one can effectively diagnose behavioral addiction using digital data traces from social media platforms. Focusing on the TikTok short-format video platform as a case study, we employ a novel mixed methodology of combining survey responses with data donations of behavioral traces. We survey 1590 TikTok users and stratify them into three addiction groups (i.e., less/moderately/highly likely addicted). Then, we obtain data donations from 107 surveyed participants. By analyzing users' data we find that, among others, highly likely addicted users spend more time watching TikTok videos and keep coming back to TikTok throughout the day, indicating a compulsion to use the platform. Finally, by using basic user engagement features, we train classifier models to identify highly likely addicted users with $F_1 \geq 0.55$. The performance of the classifier models suggests predicting addictive users solely based on their usage is rather difficult.

Figures (10)

• Figure 1: Participants' answers for TikTok usage questions across different addiction levels. The usage pattern of highly likely addicted (HLA) and less likely addicted (LLA) participants are distinct: HLA participants report to spend more time on TikTok and use it more often than LLA participants.
• Figure 2: Data collection pipeline and the number of participants at each stage.
• Figure 3: Average daily hours spent on watching videos on TikTok. Circular dots represent mean values and vertical bars represent standard errors. Triangular dots represent median values, same for later figures. HLA users spend more time on TikTok watching videos than LLA users.
• Figure 4: Session measures across addiction groups. Session duration is measured in hours. HLA and MLA users return to TikTok more often than LLA users and MLA users also have watched fewer videos each time than LLA users.
• Figure 5: Mean daily video engagement. No difference is observed in users' attention and interactions with videos across addiction groups.