The main association
Loneliness was associated with lower self-efficacy and higher internet-addiction scores among 1,323 students in Grades 7 to 9, according to a cross-sectional survey. Self-efficacy is a measure of how capable someone feels of handling demands. After controlling for gender, grade and age, the reported coefficients were -0.296 for self-efficacy and 0.247 for internet addiction, with p < .001 for both. In plain terms, students who scored higher for loneliness also tended to score lower on the study's self-efficacy measure and higher on its internet-addiction measure. The figures describe a pattern in the data, not proof that loneliness caused either outcome.
The broader correlation pattern pointed in the same direction. Loneliness and the study's sleep-quality scores were negatively correlated with self-efficacy and positively correlated with internet addiction, while internet addiction was negatively correlated with self-efficacy. Because this was a cross-sectional survey, the analysis cannot establish temporal order or cause and effect. It shows how the measures moved together in this group, not which factor caused another.
A small part of the link ran through internet addiction
The researchers then tested whether internet addiction acted as a statistical mediator, meaning that it accounted for part of the observed loneliness and self-efficacy association within the model. When loneliness and internet addiction were entered together, both remained negatively associated with self-efficacy: the coefficients were -0.273 for loneliness and -0.094 for internet addiction, with p < .001 for each. The estimated indirect effect through internet addiction was -0.023, with a 95% bootstrap confidence interval from -0.040 to -0.007. It represented 7.84% of the total effect, a small share of the overall statistical association.
That result does not turn the pathway into a causal explanation. It means the data fit a model in which higher loneliness was associated with higher internet-addiction scores, while higher internet-addiction scores were associated with lower self-efficacy when the measures were considered together. The cross-sectional design cannot show that loneliness led to internet addiction, or that internet addiction led to lower self-efficacy.
Sleep altered two statistical paths
Sleep quality changed the strength of two links in the model. It moderated the direct association between loneliness and self-efficacy, and it moderated the first-stage association between loneliness and internet addiction. It did not moderate the second-stage association between internet addiction and self-efficacy. In ordinary language, the relationship between loneliness and the other measures varied across sleep-quality levels, but the modeled internet-addiction and self-efficacy link did not show the same variation.
The sleep measure needs careful reading. The Athens Insomnia Scale kept its original scoring: higher total scores indicated poorer sleep quality, while lower scores indicated better sleep quality. The moderation finding therefore concerns the strength of associations across those score levels. It does not by itself show that better sleep protected students from loneliness or internet addiction.
What the survey can tell us
The survey produced 1,323 valid responses from 1,353 questionnaires, a 97.78% valid-response rate, through convenience recruitment among Grades 7 to 9 students. The researchers assessed loneliness, self-efficacy, internet addiction and sleep quality with questionnaire measures, then used descriptive statistics, Pearson correlations, multiple regression, mediation and moderation analyses. The moderated mediation analysis used SPSS Process Model 8 and controlled for gender, grade and age.
A common-method check found five factors above the test's cutoff of 1, and the first accounted for 23.163% of the overall variation in responses, below the recommended 40% threshold. The check provides context for the questionnaire results, but it does not change the study's observational design.
Within this surveyed group, loneliness, self-efficacy, internet addiction and sleep quality formed a linked statistical pattern, with internet addiction accounting for only a modest part of the loneliness and self-efficacy association. Because recruitment was by convenience and the design was cross-sectional, the findings cannot establish a causal mechanism or show that the pattern applies broadly beyond the sampled students.
Paper data and sources
Original title: Relationship between loneliness and self-efficacy: the mediating role of internet addiction and the moderating role of sleep quality
Authors: Li X, Sun Y
Journal/Repository: Frontiers in psychiatry
Status: Peer-reviewed
First online: 2026-10-08
DOI: Not available
Original paper · Full text