The research, published in the journal Large-scale Assessments in Education, was led by Samuel Falcon of the University of Las Palmas de Gran Canaria together with Yuney Santos of Mid Atlantic University. Rather than looking only at test scores, the team turned its attention to what the authors call psychoeducational resources: the relational, motivational, and emotional assets that determine whether a student can actually use a second chance when one is offered. The study draws on the Spanish sample of PISA 2022, the Programme for International Student Assessment, which in Spain covered 29,775 students aged 15 to 16. Of these, 5,642 students, or 18.95 percent, had repeated a grade at least once, a striking figure when set against the OECD average of around 9 percent.

Spain is an unusually important laboratory for this question. Retention there has long been more frequent than in most other wealthy countries, persisting even through reforms intended to curb it. In PISA 2022, 22 percent of Spanish 15-year-olds reported having repeated a grade after entering primary school. That means retention is not a marginal intervention affecting a small and exceptional group; it is a routine feature of the educational trajectory of a large share of Spanish adolescents. Understanding what that experience does to students, beyond their mathematics and reading scores, therefore matters for an entire generation.

The methodological challenge is that students are not randomly assigned to repeat a grade. Retained students tend to differ from their promoted peers before the decision is ever made: in the Spanish sample they came from families with lower economic, social and cultural status, had fewer years of early childhood education, were more likely to be first-generation immigrants, more likely to be male, and more likely to have been born late in the year, a well-known relative-age effect. A naive comparison of retained and non-retained students would therefore confound the effect of retention with these pre-existing differences. To address this, the researchers used propensity score matching, a statistical technique that estimates each student’s probability of being retained from observed background characteristics and then pairs each retained student with a non-retained student of similar probability.

The matching procedure was unusually careful. The team tested several specifications, including generalised full matching and nearest-neighbour matching at 1:1 and 1:2 ratios with various calipers, and settled on 1:1 nearest-neighbour matching without replacement, which kept every retained student in the analysis while achieving excellent balance. After matching, the mean of the maximum absolute standardised mean differences was 0.026 for the original covariates, well below the conventional 0.10 threshold, and 0.074 even for higher-order and interaction terms. Missing values in the background covariates were handled through multiple imputation across five datasets, with estimates pooled using Rubin’s rules. The analyses also incorporated PISA’s complex survey design, combining matching weights with the final student weights and 80 replicate weights using balanced repeated replication with Fay’s adjustment.

The seven outcomes spanned three domains. School integration was captured by sense of belonging and perceived quality of student-teacher relationships. Learning-related resources included self-directed learning self-efficacy, perseverance, and curiosity. Emotional functioning was measured through emotional control and stress resistance, indices drawn from the OECD framework on social and emotional skills. The results were strikingly consistent in direction. After matching and adjustment, retained students scored significantly lower on six of the seven resources, with associations surviving a Benjamini-Hochberg false discovery rate correction across outcomes.

The largest differences appeared in the learning-related domain. Self-directed learning self-efficacy, students’ confidence in managing their own learning tasks independently, showed the biggest gap, with an effect size of d = -0.27, followed by perseverance at d = -0.25. Curiosity was lower as well (d = -0.12), while the relational and emotional differences were smaller: sense of belonging (d = -0.07), student-teacher relationships (d = -0.09), and emotional control (d = -0.09). Only stress resistance showed essentially no difference, with an effect size of -0.01 and a regression coefficient of exactly zero that was nowhere near statistical significance. Sensitivity analyses using alternative matching specifications reproduced the same pattern, suggesting the findings are not an artefact of one particular matching choice.

There is a deep irony in the two largest effects. The entire rationale for retention is that an extra year gives students time to master content they could not grasp the first time. But that rationale presupposes students who feel capable of organising their own learning, who persist when tasks become difficult, and who remain open to new ideas. The study suggests retained students are precisely the ones who report being weakest in these resources. The authors interpret this as consistent with retention being embedded in a longer history of academic struggle, in which the relationship with effort and academic agency may already be fragile. Repeating a grade, in this light, may not rebuild that relationship; it may simply extend a trajectory in which students feel less and less competent at steering their own learning.

The relational findings matter for a similar reason. Retained students are typically separated from their original peer group, placed with younger classmates, and carry a visible marker of prior failure in front of teachers and peers. Cross-national PISA evidence has repeatedly shown that repeaters report lower belonging, and the Spanish data confirm this pattern even after matching on background characteristics. Interestingly, the picture on emotions was mixed: retained students reported more difficulty regulating irritation, anger, and frustration, yet showed no difference in general stress resistance. The authors suggest these two constructs capture distinct aspects of emotional functioning, and that retention may be more closely tied to emotional regulation in challenging situations than to a general tolerance for pressure, though they caution that this interpretation requires longitudinal testing.

The researchers are careful about what their design can and cannot show. Because the data are cross-sectional and observational, the results are adjusted associations, not causal effects. The propensity score model could not account for prior achievement, attendance, learning difficulties, behaviour, teacher judgements, or the school-level practices that shape retention decisions, so residual confounding is likely. There is also no way to know whether retained students already had lower belonging or perseverance before repeating, or whether the differences emerged afterwards. Missing data posed a further complication, particularly for self-directed learning self-efficacy, where questionnaire routing tied to COVID-19 school closures left 34 percent of observations unavailable, and missingness was consistently higher among retained students. Even so, the consistency of the pattern across six outcomes, three domains, and multiple matching specifications is difficult to dismiss.

The policy implications are sobering. If repeating a grade is meant to be a second opportunity, the study suggests that time alone does not create the conditions for recovery. The authors argue that retention should not be treated as an intervention in itself, and they point to candidate domains for future support research: transition planning after retention, monitoring of school belonging, teacher guidance aimed at rebuilding academic confidence, and structured help for self-regulated learning, with particular attention to self-directed learning self-efficacy and perseverance, the two resources showing the largest gaps. More broadly, the findings reinforce a growing argument that evaluations of grade retention should not be restricted to test scores. Whether students remain connected to school, confident in managing their learning, and emotionally equipped to continue may ultimately determine whether the extra year helps or harms, and those dimensions are invisible in achievement data alone. For a country where nearly one in five adolescents repeats a grade, that is a message worth taking seriously.

Subject of Research: Associations between grade retention and psychoeducational resources among Spanish students in PISA 2022

Article Title: Grade retention and psychoeducational resources: evidence from PISA 2022 using propensity score matching

Article References: Falcon, S., & Santos, Y. (2026). Grade retention and psychoeducational resources: evidence from PISA 2022 using propensity score matching. Large-scale Assessments in Education, 14(1), Article 48. https://doi.org/10.1186/s40536-026-00323-9

Image Credits: AI Generated

DOI: 10.1186/s40536-026-00323-9

Keywords: grade retention, PISA 2022, propensity score matching, psychoeducational resources, school belonging, student-teacher relationships, self-directed learning self-efficacy, perseverance, curiosity, emotional control, Spain, educational psychology