A Meta-Analytical Review of the Determinants of Social Media Discontinuance Intentions

ABSTRACT

The use of social media has grown tremendously, but a considerable number of individuals have stopped using it. This meta-analysis aims to examine the factors that contribute to discontinuing social media use by reviewing 88 studies with a cumulative sample size of 42,159, including 33 effect sizes. Our study reveals that various stressors, including messaging overload (CO), social overload (SO), information collection overload (IO), system feature overload (SFO), privacy concerns (PC), and negative emotions such as technostress, fatigue (SNF), guilt, and dissatisfaction, are significantly correlated with social media discontinuance (DUIN). It is worth noting that only gratifications were negatively associated with both discontinuance and fatigue, but not with all other inhibitors. Furthermore, self-disclosure (S-disc), social comparison (SC), and fear of missing out (FoMo), as well as addiction, were significantly associated only with fatigue. Theoretical and practical implications are discussed, emphasizing the importance of social media operators balancing content supply with actual user demand to prevent overload, negative emotions, and discontinuance.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15205436.2023.2263872.

Notes

¹ If the primary studies report demographics as effect sizes, they will be tested as such. However, if the primary studies report demographics as sample characteristics rather than effect sizes, they will be tested as moderators.

² To address the inconsistency in the number of points on the scales, the means reported were adjusted by subtracting the medians of the corresponding scales. For instance, if a study employed a 7-point scale and reported an average income level of 3, the adjusted mean would be −1 (calculated as 3 minus 4).

³ Since most studies were conducted in China, Korea, Japan, and Taiwan (52%), we classified these regions as Confucian culture. Zhang et al. (2005), whereas the rest of the regions were termed non-Confucian cultures.

⁴ Since most studies studied social networking sites (SNS) (70%), we classified social media platforms as social network sites vs. other forms of social media. For a definition of social media and social network sites, see Boyd and Ellison (2007), Feng et al. (2019), and Liu et al. (2018)

⁵ Chu and Kim (2011).

⁶ Some variables that are conceptually similar were unified into the same name. For instance, guilt, shame, and regret are combined according to Kugler and Jones (1992).

Additional information

Funding

This work was supported by the Hong Kong Baptist University [Start-up grant]; National Social Science Fund of China [18BXW082].

Notes on contributors

Guangchao Charles Feng

Guangchao Charles Feng, PhD is an associate professor at School of Communication, Hong Kong Baptist University. His main research areas are research methods, new media studies, computational advertising, message effects, and political communication.

Xianglin Su

Xianglin Su and Yiru He are both master’s students under the supervision of Dr. Feng and were primarily involved in the data collection and visualization aspects of this research.