Narcissism and Problematic Internet Use: Differential Links to Internet Versus Social Media Addiction—A Multidimensional Meta-Analysis

Problematic Internet use (PIU) is prevalent and harmful, yet findings regarding narcissism remain inconsistent. We meta-analyzed 143 effects from 74 studies ( N = 68,705) using random-effects models and relative weight analyses to examine global, dual-dimensional, and trifurcated narcissism across PIU domains. Global narcissism was positively associated with PIU ( r = 0.26, 95 percent confidence interval (CI) 0.22, 0.30) and remained significant after trim-and-fill, selection model, and outlier analyses. Subgroup analyses showed a positive association for social media addiction ( r = 0.26, 95 percent CI 0.21, 0.30), whereas the Internet addiction subgroup showed a smaller but still significant association ( r = 0.16, 95 percent CI 0.07, 0.26). Given the substantial heterogeneity across pooled outcomes, these findings should be interpreted as indicating an overall positive pattern rather than uniform consistency across studies. Dimensional analyses indicated that estimated associations varied according to the dimensional composition of narcissism measures. In dual-dimensional models, grandiose and vulnerable narcissism showed similar estimated associations with general PIU ( r = 0.20–0.24) and problematic social media use ( r = 0.21–0.22). For Internet addiction, however, the estimated association increased from r = 0.09–0.47 as vulnerable narcissism increased and decreased from r = 0.47–0.09 as grandiose narcissism increased, suggesting a stronger estimated role for vulnerability in this comparatively small subgroup. In trifurcated models, interpersonal antagonism showed the strongest positive estimated association across PIU domains, whereas narcissistic neuroticism showed inverse estimated associations in several analyses, and agentic extraversion showed minimal or nonsignificant effects. Conventional study-level moderators were not statistically significant, but relative-weight-based dimensional composition analyses suggested that estimated associations varied as a function of dimensional composition. Overall, these findings suggest that antagonistic and vulnerable aspects of narcissism may be especially relevant to PIU while also highlighting the importance of moving beyond global narcissism scores and interpreting dimensional estimates cautiously.