Gender Differences in Loneliness Across the Lifespan: A Meta-Analysis


Marlies Maes, KU Leuven, Research Foundation Flanders, Belgium
Pamela Qualter, University of Manchester, UK
Janne Vanhalst, Ghent University, Belgium
Wim van den Noortgate, KU Leuven, Belgium
Luc Goossens, KU Leuven, Belgium


Loneliness, which is the unpleasant feeling that occurs when people perceive their network of social relationships to be deficient in quantitative or qualitative ways, has been related to a broad range of mental and physical health problems. Many studies have examined whether gender represents a vulnerability factor for loneliness, but results have been largely inconsistent, and no consensus has been reached. Therefore, we aimed to synthesize the available evidence on gender differences in loneliness. In addition to examining that global effect, we investigated the moderating effects of participants’ age, types of loneliness, the country in which the study was conducted, the socioeconomic, ethnic, and clinical status of the participants, the geographical representation of the sample, and the year in which the study was published.


A systematic literature was conducted and revealed, after applying the selection criteria, 751 effect sizes from 638 studies. A total of 399,798 individuals were included in the present meta-analysis (45.56% male). Several studies reported on multiple effect sizes, so to account for possible dependency among effect sizes, we conducted multilevel meta-analysis. Specifically, we conducted a cross-classified three-level model with random sampling variance (Level 1) and within-study variance (Level 2). At Level 3, we considered two sources of random variation, that is, between-study variance and between-instrument variance. Analyses were conducted with the metafor package (Version 1.9–9) in R. At the Open Science Framework, both the dataset ( and analysis scripts ( are available.


First, when focusing on the 544 effects for which sufficient information was available to calculate a standardized mean difference, we found a close-to-zero mean effect of g = 0.08 (SE = 0.03, p = .005, 95% CI [0.02, 0.13]), suggesting that males are slightly lonelier than females. Second, when the analysis was based on all 751 effects and, thus, also included the effects for which we had to make assumptions (e.g., assuming equal sample sizes when only a total sample size was provided), we obtained a similar effect of g = 0.07 (SE = 0.02, p = .003, and 95% CI [0.03, 0.12]). Third, we focused on the effect sizes derived from the larger samples with a minimum of 100 male and 100 female participants. This analysis, based on 376 effects, yielded a non-significant mean effect size of g = 0.04 (SE = 0.02, p = .078 and 95% CI [-0.01, 0.09]). Most moderators did not significantly predict gender differences in loneliness, except for age group, sampling area, and year of publication. All effects, however, were small.


We found very similar mean levels of loneliness for males and females, from childhood through old age, for different types of loneliness, and across a range of demographic background variables, suggesting that males and females are more alike than they are different on self-reported loneliness.


Link to paper > Contact > Website >
Back to Spotlights