Girls From Educated Families Faced Surprising Loneliness Spike During Pandemic School Closures

When classrooms around the world went dark in 2020 and 2021, an entire generation of adolescents traded hallway conversations, lunchtime banter, and after-school rituals for screens and silence. Public health researchers have long warned that loneliness among teenagers is not a trivial emotional state but a robust predictor of depression, anxiety, self-harm, and diminished lifelong health. A new cross-national study published in the International Journal for Equity in Health now offers one of the clearest pictures yet of who suffered most during pandemic-related school closures, and its central finding defies a common assumption: adolescents from highly educated families were not shielded from loneliness, and in one striking subgroup, they fared worse.

The research team, led by Yuko Inoue of the University of Tokyo and Institute of Science Tokyo, together with Rebecca Wilson of the University of Bristol, Naoko Adachi, and corresponding author Shiho Kino, drew on an unusually powerful data source: the 2022 Programme for International Student Assessment, or PISA, administered by the Organisation for Economic Co-operation and Development. PISA is best known for benchmarking the academic performance of fifteen-year-olds, but the 2022 cycle included a retrospective question asking students whether they had felt lonely during the pandemic-related school closures. By harvesting responses from 91,591 students across 35 OECD countries, the investigators assembled a dataset large and geographically broad enough to detect patterns that smaller, single-country studies could easily miss.

The headline number is sobering on its own. Overall, 42.8 percent of students agreed or strongly agreed that they had felt lonely while schools were shut, and 14.3 percent strongly agreed. In other words, roughly two in five adolescents across some of the world’s wealthiest democracies reported experiencing loneliness during a period when the school, for most of them the primary arena of social life, had effectively disappeared. Because the question was retrospective, asked in 2022 about the closure period, the authors are careful to frame the measure as perceived loneliness during that time rather than a real-time clinical diagnosis. Still, the sheer scale of agreement across such a diverse sample underscores how disruptive the closures were to adolescent social connectedness.

Methodologically, the study goes well beyond simple percentages. The team used multilevel ordinal logistic regression, a statistical framework appropriate when the outcome, loneliness, is measured on an ordered scale and when students are nested within schools and countries, allowing the model to account for clustering at each level. The central exposure of interest was parental education, used as a proxy for socioeconomic status, with the highest parental attainment categorized and tertiary education, meaning university-level degrees, serving as the reference point for the most advantaged group. The models adjusted for an unusually rich set of covariates: the duration of school closures, students’ perceived learning during closures, difficulties with self-directed learning, self-efficacy in self-directed learning, and measures of family and school support. This adjustment strategy matters because these factors plausibly lie on the pathway between socioeconomic background and loneliness, and ignoring them could distort the estimated associations.

The first major result concerns gender. Female students had significantly higher odds of reporting greater loneliness than male students, with an odds ratio of 1.80 and a 95 percent confidence interval of 1.66 to 1.95. An odds ratio of this magnitude, and with a confidence interval that does not come close to crossing 1.0, indicates a robust and substantial difference: girls were nearly twice as likely as boys to report elevated loneliness during the closures. This aligns with a broader literature suggesting that adolescent girls tend to derive more of their emotional support from peer networks and may be more sensitive to interruptions in face-to-face contact, although the PISA data cannot disentangle the underlying psychological mechanisms.

The second and more surprising result emerges from the interaction between gender and parental education. In epidemiological terms, an interaction tests whether the effect of one factor depends on the level of another. Here, the researchers found a statistically significant interaction, with particularly elevated odds of loneliness among female students whose parents had attained tertiary education. The odds ratio for the interaction term was 1.35, with a 95 percent confidence interval of 1.11 to 1.65. Translated into plain language, the protective effect that higher socioeconomic status is often assumed to confer on adolescent mental health did not extend to loneliness during the closures, and for girls from university-educated families the burden was actually amplified relative to what the main effects alone would predict.

Why might this counterintuitive pattern arise? The authors themselves are measured, noting that adolescents with tertiary-educated parents are not necessarily protected from loneliness. Several plausible explanations circulate in the wider literature. Families with more resources may have maintained academic continuity through devices, tutors, and parental supervision, but academic support is not the same as peer connection. Highly educated parents were also more likely to work in jobs compatible with strict remote work, potentially meaning tighter household isolation rather than looser rules. Adolescent girls in such households may have experienced heightened pressure around self-directed learning and achievement, and the loss of structured social time may have hit hardest among those most embedded in school-based friendship networks. The PISA data cannot adjudicate among these hypotheses, but the interaction finding signals that policymakers should not assume socioeconomic advantage automatically buffers emotional wellbeing during educational disruptions.

The study’s limitations deserve honest treatment. The loneliness measure is a single self-report item, which is efficient for cross-national surveys but cannot capture the depth, duration, or clinical significance of loneliness. Retrospective reporting introduces recall bias, since students answered in 2022 about experiences that had occurred up to two years earlier, and mood at the time of reporting may color memories of the closure period. The cross-sectional design precludes causal inference: the associations describe who reported loneliness, not what caused it. Parental education, while a widely used and defensible proxy, is only one dimension of socioeconomic status and may behave differently across the 35 countries in the sample. Finally, the data cover OECD countries only, so the findings cannot be generalized to low- and middle-income settings where school closures often lasted longer and support infrastructures were thinner.

Even with those caveats, the implications are concrete. The authors conclude that emotional and academic support must be provided across gender and socioeconomic groups during educational disruptions, not targeted narrowly at disadvantaged households. Schools planning for future closures, whether from pandemics, natural disasters, or other emergencies, should build in structured social contact, monitor loneliness as systematically as they monitor learning loss, and pay particular attention to girls, who bore a consistently heavier emotional load in this data. The work was supported by grants from the Japan Society for the Promotion of Science, including a Grant-in-Aid for JSPS Fellows (24KJ0065) and a Grant-in-Aid for Early-Career Scientists (22K17266), and the authors report no competing interests. As a secondary analysis of publicly available OECD data, the study required no new ethical approval. What it delivers is a warning worth heeding: when schools close, loneliness does not discriminate by parental diploma, and the adolescents who seem safest on paper may be quietly struggling the most.

Subject of Research: Adolescent loneliness and socioeconomic and gender differences during COVID-19 pandemic-related school closures

Article Title: Adolescent loneliness and socioeconomic factors during the pandemic-related school closures: a cross-sectional study using retrospective PISA data

Article References: Inoue, Y., Wilson, R., Adachi, N., & Kino, S. (2026). Adolescent loneliness and socioeconomic factors during the pandemic-related school closures: a cross-sectional study using retrospective PISA data. International Journal for Equity in Health. https://doi.org/10.1186/s12939-026-03018-3

Image Credits: AI Generated

DOI: 10.1186/s12939-026-03018-3

Keywords: adolescent loneliness, school closures, COVID-19, PISA 2022, socioeconomic status, parental education, gender differences, mental health, OECD countries, public health, self-directed learning, health equity