Higher daily social media use was consistently associated with self-reported depression among US adults across 11 independent repeated cross-sectional samples, according to a study published in npj Mental Health Research.
Researchers analyzed annual Media Behavior & Influence Study data collected from 2014 to 2024. The analytic sample included approximately 183,300 adults aged 18 to 70 years, with 15,410 to 17,869 participants per year. Samples were balanced to US Census benchmarks for age and gender.
Self-reported depression was assessed using a single yes-or-no survey item rather than a validated multi-item depression scale. Daily social media use was estimated from participants’ reported probability of engaging in social media during each hour of a typical day.
Researchers used covariate-adjusted logistic regression to examine associations between media use and self-reported depression and balanced random forest models to classify self-reported depression using demographic and media-use variables. Model discrimination was assessed using area under the receiver operating characteristic curve, and feature importance was assessed using Gini, permutation, and SHAP analyses.
Social media use was positively associated with self-reported depression in all 11 annual samples after adjustment for demographic covariates. An additional 60 minutes of daily social media use corresponded to approximately 1.17 times the odds of self-reported depression. Descriptive analyses also showed a positive gradient in self-reported depression prevalence across increasing quantiles of social media use in each study year.
Balanced random forest models yielded area under the receiver operating characteristic curve values ranging from 0.66 to 0.74, indicating moderate discrimination, and accuracy ranging from 72% to 77%. When Gini, permutation, and SHAP rankings were considered together across the study period, personal income ranked first overall, and social media use ranked second.
Models trained using pooled 2014 to 2018 data and tested on 2019 to 2024 samples produced area under the receiver operating characteristic curve values ranging from 0.68 to 0.72. SHAP analyses indicated that social media use became a positive contributor to model-classified depression at approximately 150 minutes of daily use across several study years. “The 150-min value should therefore be treated as a hypothesis for future longitudinal or experimental studies,” wrote co-first author Martin L. Block of the Integrated Marketing Communications, Medill School of Journalism, Northwestern University in Evanston, Illinois, and colleagues.
The approximately 150-minute point was not a clinical cutoff, and the cross-sectional analysis could not establish the direction of the association.
Limitations included the use of a commercial data set and a single self-reported depression measure rather than a validated clinical scale, which may have resulted in misclassification. The recruitment methods may have introduced selection bias. Conducting the surveys at the same time each year may have introduced unmeasured confounding. The cross-sectional design prevented causal or directional conclusions about the association between social media use and depression. The sample was limited to adults aged 18 to 70 years, and the survey did not distinguish passive from active social media use. The researchers noted that feature-importance measures may reflect variability, distributional characteristics, and measurement properties rather than etiologic relevance or causal strength.
Funding was provided in part to Hans C. Breiter by the Office of Naval Research and a Jim Goetz donation to the University of Cincinnati College of Engineering and Applied Science. The researchers declared no competing interests.
Source: npj Mental Health Research
