Examining the relationship between problematic social media use and caffeine use disorder in the context of physical activity: a cross-sectional moderation study among university students

Introduction Problematic social media use (PSMU) and caffeine use disorder (CUD), a behavioral pattern without formal diagnostic status and a DSM-5 condition listed for further study, respectively, share phenomenological features (tolerance, withdrawal, compulsive engagement) that make their statistical co-occurrence among young adults theoretically plausible and warrant joint empirical examination. Yet the potential moderating role of lifestyle factors such as physical activity remains largely unexplored. This study examined the relationship between problematic social media use, hereafter termed social media addiction (SMA), and caffeine use disorder (CUD) among university students, and determined whether physical activity (PA) level moderates this relationship. Methods A cross-sectional correlational survey design was employed. Data were collected online from 408 Turkish university students (59.1% women; M = 22.38 years, SD = 2.14) using the Bergen Social Media Addiction Scale (BSMAS), the Caffeine Use Disorder Questionnaire (CUDQ), and the International Physical Activity Questionnaire-Short Form (IPAQ-SF). Pearson correlation, one-way ANOVA with Tukey HSD and Games-Howell post hoc comparisons, and hierarchical moderated regression were employed. Results A statistically significant small-to-moderate positive correlation emerged between SMA and CUD ( r = 0.374, p < 0.001). SMA scores did not differ significantly across PA groups (inactive, minimally active, sufficiently active), F (2, 405) = 0.177, p = 0.838, η 2 p = 0.001. CUD scores, however, differed significantly by PA level, F (2, 405) = 4.458, p = 0.012, η 2 p = 0.021, with inactive students reporting lower CUD than both active groups, a pattern that is descriptively compatible with, but does not directly test, the ergogenic caffeine use hypothesis. Physical activity did not moderate the SMA-CUD relationship ( β = 0.043, p = 0.358; Δ R 2 = 0.002), although SMA remained a significant direct predictor of CUD ( β = 0.373, p < 0.001). Discussion These cross-sectional findings suggest that PSMU and CUD co-occur to a modest but detectable degree among university students, and that habitual PA neither attenuates nor amplifies this association. Given the modest effect sizes and the design, implications for screening and intervention are discussed cautiously and treated as hypothesis-generating for prospective and experimental work.