Schools Raise Digital Skills but Fail to Close the Socioeconomic Gap, 33-Country Study Finds

A sweeping analysis of data from 33 countries has found that while schools can lift students’ digital competence overall, they do not appear to narrow the gap between children from wealthier and poorer families. The study, published in Large-Scale Assessments in Education, drew on the International Computer and Information Literacy Study (ICILS) 2023 and combined country-specific multilevel statistical models with a two-step meta-analytic synthesis to ask a deceptively simple question: does digitalization in schools actually reduce socioeconomic inequality in digital skills? The answer, according to the researchers led by Tamara Kastorff, is a sobering no.

The starting point of the analysis was the strength of the association between socioeconomic status, or SES, and digital competence. Across all 33 participating education systems, the pooled estimate showed that a one-unit increase in SES was associated with a gain of 17.73 points on the digital competence scale (standard error 0.93, 95 percent confidence interval from 15.84 to 19.61). In practical terms, students from higher-SES backgrounds scored substantially higher on the performance-based assessment of computer and information literacy than their lower-SES peers. Crucially, this association was positive in every single country examined, although its magnitude varied widely, from a low of 7.11 points to a high of 28.09 points per SES unit.

That variation between countries was itself striking. The between-country variance was estimated at 24.41, corresponding to an I-squared value of 90.18 percent, meaning that nearly all of the observed variability in effect sizes reflected genuine differences between national contexts rather than statistical noise. The heterogeneity test was highly significant, with Q(32) equal to 295.70 and p below .001. In other words, socioeconomic inequality in digital competence is a universal phenomenon, but its steepness depends on where a child happens to grow up, a pattern the authors describe as underscoring the structural nature of the digital divide.

The researchers then turned to what schools are actually doing with technology, and here the picture became more nuanced. Several classroom-level practices were positively associated with students’ digital competence. Learning about internet-related tasks in school showed a positive pooled effect of 1.77 points. General classroom applications of information and communication technology showed a strong positive association of 6.72 points, and teachers’ use of general utility software was associated with a gain of 3.67 points. Inquiry-based teaching, as reported by teachers, added a positive effect of 1.86 points, and teachers’ use of ICT for basic classroom activities contributed 1.51 points.

Not every form of technology use was beneficial, however. Students’ use of specialist classroom applications was negatively associated with digital competence, with a pooled effect of minus 8.18 points. Teachers’ use of digital learning tools showed a negative association of minus 4.55 points, and experiential scenario-based teaching was also negatively linked, at minus 1.89 points. The authors offer two possible interpretations, which the cross-sectional data cannot disentangle. One draws on cognitive load theory and research on instructional guidance: complex or minimally guided activities may place heavy demands on learners, particularly when prior knowledge differs widely. The other concerns selection: specialist applications and digital learning tools may simply be used more often in remedial settings with lower-achieving students, so the negative associations could reflect who receives such instruction rather than harm caused by the tools themselves.

Perhaps the most telling finding concerned school leadership. None of the principal-reported characteristics, including principals’ views on using ICT for educational outcomes, their expectations of teachers’ ICT use, their priorities for facilitating technology, or their own use of ICT for administration and communication, showed a statistically significant association with students’ digital competence. These analyses were based on a reduced subset of 21 education systems because not all countries administered the principal questionnaires. The pattern suggests that what matters is how digital media is embedded in everyday teaching and learning, not whether schools formally endorse digitalization at the policy level.

The central question, however, was whether any of these school-level factors could moderate the link between SES and digital competence, and here the results were largely null. Across student-reported indicators such as learning internet-related tasks, general classroom ICT applications, and specialist applications, no significant moderating effects emerged, and the between-country heterogeneity of these moderation estimates was essentially zero. The same held for all teacher-reported practices: inquiry-based teaching, scenario-based teaching, basic and advanced ICT use, digital learning tools, and utility software all failed to significantly change the SES gradient. The benefits of good digital instruction appear to accrue roughly equally across socioeconomic groups, raising overall competence levels while leaving relative gaps untouched.

There was one small exception. Principals’ use of ICT for general school-related activities showed a statistically significant negative moderating effect of minus 0.14 points (95 percent confidence interval from minus 0.25 to minus 0.03), with no heterogeneity across the 21 education systems analyzed. This suggests that in schools where principals reported higher levels of ICT use for general administrative purposes, the association between SES and digital competence was slightly weaker. The authors interpret this cautiously as a possible hint that broader institutional digital maturity is linked to marginally weaker disparities, but they emphasize that the isolated nature of this single effect warrants restraint.

The authors situate their findings against earlier work that had reported compensatory effects of school digital strategies. One prior study found that inquiry-based digital teaching was associated with weaker SES differences in students’ self-perceived digital competence, but also that obligatory digital strategies could reinforce inequality because higher-SES students benefited more. That work relied on self-reported outcomes, whereas the present study used performance-based assessments, and the lack of consistent moderation here suggests that school-level digital strategies may boost confidence without translating into reduced disparities in demonstrated skill. Complementary evidence indicates that compensatory effects may occur in specific national contexts, pointing to the possibility that such mechanisms are context-dependent rather than systematic.

The study has clear limitations that the authors acknowledge. The data are cross-sectional, so causal direction cannot be established, and reverse causality between instructional practices and competence cannot be ruled out. The analyses did not incorporate country-level indicators such as national digitalization policies, funding structures, or broadband access, despite the substantial between-country heterogeneity, and they did not capture detailed home learning environments beyond SES, such as parental digital capital or informal learning opportunities. School-level measures were aggregated from student, teacher, and principal reports, which may not fully reflect the quality of digital instruction. The authors conclude that if digital competence is to serve as a lever for educational equity, general digitalization strategies alone appear insufficient; targeted measures such as structured digital training programs, differentiated support for students with limited digital resources, and funding mechanisms that prioritize disadvantaged schools may be required to turn classroom technology use into genuinely narrower gaps.

Subject of Research: Socioeconomic disparities in students' digital competence and the role of school-level digitalization across 33 countries

Article Title: Do schools narrow the socioeconomic digital divide? Meta-analytic evidence from ICILS 2023

Article References: Kastorff, T., Scherer, R., Müller, M., Booij, G., Kennedy, A., Strietholt, R., & Greiff, S. (2026). Do schools narrow the socioeconomic digital divide? Meta-analytic evidence from ICILS 2023. Large-scale Assessments in Education, 14(1), Article 57. https://doi.org/10.1186/s40536-026-00330-w

Image Credits: AI Generated

DOI: 10.1186/s40536-026-00330-w

Keywords: digital divide, digital competence, socioeconomic status, ICILS 2023, educational equity, ICT in schools, meta-analysis, multilevel modeling, classroom technology, school digitalization, large-scale assessment, digital literacy