Chronic school absenteeism

1. Introduction: an indicator that took centre stage

The vignette does not prove a fact: it shows the gesture under suspicion. After the Every Student Succeeds Act (ESSA) of 2015, 36 states of the United States adopted chronic absenteeism as an indicator of quality and of student success (Henderson & Fantuzzo, 2023). The promise was to detect, from registers already on file, who will have lower achievement if the absences continue. That use put the threshold centre stage, even though the assumption that every absence informs risk equally had not been tested.

As a conceptual framework, Childs and Lofton (2021) call it a wicked problem. Concentrating policy on group size, teaching, school choice or leadership distracts from stacked causes and works as a mask: the student who is absent is the one those reforms reach least. What is needed is a framework across the whole system and in other public sectors.

Descriptive empirical finding. Dee (2024) shows, with state data, that the rate rose on the return to in-person schooling and that academic recovery depends in part on attending. The next section separates that description from virtual instruction. The question is explicit: what is known, between 2021 and 2026, about scale, the validity of the indicator, causes and the effectiveness of responses, and with what quality of evidence? The text separates empirical finding, conceptual framework and inference, and it does not turn an association into a cause, nor a title without an abstract into a result.

2. The scale of the problem after the pandemic

Descriptive empirical finding. Between 2018–2019 and 2021–2022, the share of students in United States public schools with chronic absenteeism rose by 13.5 percentage points. Dee (2024) describes it as a 91 % increase, some 6.5 million additional students. The datum is at state level and does not explain the jump. The increases are positively associated with the prevalence of closures in 2020–2021: an association between aggregates, not an effect of having closed. In that analysis they are not associated with enrolment loss, COVID-19 rates, mask policies or the deterioration of youth mental health. Those nulls rule out several simple stories without consecrating closures as the sole cause.

Evans et al. (2024) observe the same increase of 13.5 points in districts: 11 017 districts, from kindergarten through grade 12, and 22 034 observations. The mean moved from 15.9 % (standard deviation 8.1 %) to 29.4 % (standard deviation 13.2 %). These are not the microdata of Dee (2024). The jump does not depend on a single aggregation, and dispersion across districts grew with the mean: a reading of the reported standard deviations, not a formal test of variances.

Association, not a causal effect. Evans et al. (2024) relate the modality of 2020–2021 to the rate of 2021–2022 in a cross-section with controls, district and year fixed effects, enrolment weighting and clustered errors. 100 % virtual instruction was associated with 6.9 points more (95 % confidence interval: 4.8–8.9) than in-person instruction; hybrid instruction, with none. In the highest-poverty quintile the difference was 10.6 points (7.2–14.1). Inference: those figures do not by themselves explain the jump of 13.5.

Outside the United States there is no comparable figure here. Graham et al. (2026) state, as context for their review and not as an estimate of this article, that after the pandemic attendance fell in many countries and remains below the previous level. The abstract lists neither countries nor magnitudes.

3. Method: critical narrative review with a screening flow

The review is narrative and critical. There is no meta-analysis of our own: variances are not pooled and no new mean effect is calculated. The base was the Crossref REST API: 14 seed queries of up to 60 rows, journal articles from 2021, and a lexical filter for absenteeism, attendance, truancy or school refusal. Screening, reproducible with screen.py, closed with manual selection and without double coding. Another reviewer could dispute some of the 24 inclusions.

There were 407 unique records from 2021 to 2026; none was removed by year. Removed were 5 corrections or comments, 41 articles from journals of low editorial assurance, 11 unrelated to school attendance (health, work, worship or meetings) and 33 out of scope for other reasons (epidemiological surveillance, hygiene, specific diseases, higher education and teacher absenteeism). That left 317. Selection gave priority to reviews, school-level trials, regression discontinuity and difference-in-differences designs, longitudinal and cross-sectional studies, measurement and conceptual texts, with priority to compulsory schooling and early childhood.

The seed, in English, did not retrieve Latin American studies in Spanish. Twelve queries of 8 rows added one record: Razeto Pavez (2026), in Perfiles Educativos, with the DOI verified in Crossref. The review includes 24 texts (23 from the seed and 1 supplementary). Nineteen have an abstract in Crossref or OpenAlex; five circulate by title only. Two are Latin American, one is Mexican, and only one, with an abstract, comes from an Ibero-American venue. Those counts describe this corpus, not regional output. Table 1 fixes the flow.

Screening stage	Count
Seed queries in Crossref (up to 60 rows)	14
Unique verified records, 2021–2026	407
Excluded: corrections, comments or replies	5
Excluded: journals of low editorial assurance	41
Excluded: do not address school attendance	11
Excluded: out of scope for other reasons	33
Eligible after screening of the seed	317
Supplementary queries (8 rows each)	12
Records added in the supplementary search	1
Included by manual selection (23 from the seed and 1 supplementary)	24
With an abstract in Crossref or OpenAlex	19
Title only, with no usable abstract	5

4. What the 10% threshold measures and what it leaves out

Chronic absenteeism, in the practice the corpus documents, is missing at least 10 % of the days enrolled. Kjolhede et al. (2025) use the federal and state categories: not at risk, from 0 % to 4.99 %; at risk, from 5 % to 9.99 %; chronic absenteeism, 10 % or more. The scale orders volume, not reason or day. Mac Iver et al. (2022) use the same cut — an attendance rate of 0.90 — to decide who receives a letter. The 10 % is a threshold of policy and of research at once.

Empirical finding. Henderson and Fantuzzo (2023) test, with ROC curves and data from a large urban district, whether all absences diagnose equally the risk of future achievement in the early grades. Only unexcused absences have precision. Excused absences do not contribute and, if they are added, dilute the signal. It is, according to the authors, the first examination of that assumption. It does not say that an excused absence is harmless for learning: it says that, for classifying risk at that stage, treating it as equivalent worsens the instrument. The vignette illustrates the problem; it does not replicate that district.

Kearney et al. (2026) do not validate a scale. They argue that attendance data have been questioned on measurement, use, context and interoperability, with little comparability across disciplines and countries. They propose a common next-generation measurement, brief and modifiable: definition, achievement and engagement, functional impairment, interoperability across sources and agencies, and ease of application. It is a roadmap. It should meet psychometric standards and be acceptable to distinct educational, scientific, legal, economic and political views. Design inference: until such a measure exists, the 10 % is a cheap substitute for comparability that remains uncertain.

Childs and Lofton (2021) call that visibility a mask. Graham et al. (2026) show the same problem in the outcome: of 29 studies, 16 report absenteeism, 4 dropout and 9 both, with heterogeneous definitions, windows and reports, and little detail on how the interventions were developed. More than half are at high risk of bias (RoB 2), and there was no meta-analysis. Inference: “reducing chronic absenteeism” is not yet a single outcome.

5. Why students miss school: an ecological map of causes

Graham et al. (2026) order the interventions with Bronfenbrenner’s ecological theory: most are multicomponent and touch several domains. Figure 1 gathers factors that the corpus documents. It is a scheme, not an estimate, and it does not turn an association, an argument or a title without results into causality.

Ecological level	Factors documented in the corpus	Source
Student and health	Illness, domestic circumstances, holidays and unexcused absences, associated with achievement; school refusal versus truancy; kindergarten health only as an object of study	Klein et al. (2022); Pérez-Marco et al. (2025); Steyer et al. (2026)
Family and resources	A small association of socioeconomic status, clearer in the family and in financial resources; family vulnerability; parental engagement with a small mean effect	Sosu et al. (2021); Razeto Pavez (2026); Strelitz et al. (2026)
School: discipline, climate and relationships	Suspension with a subsequent negative effect; a restorative package that is clear on discipline and marginal on absenteeism; closeness that buffers ethnoracial mismatch; the school as a relational unit	Singer (2025); Grant et al. (2025); Chhikara (2025); Razeto Pavez (2026)
Community and territory	Community vulnerability; a non-causal association of the rural school-based health centre; an association of fully virtual instruction with more absenteeism the following year	Razeto Pavez (2026); Kjolhede et al. (2025); Evans et al. (2024)
Policy and measurement	An ESSA (2015) indicator that does not diagnose equally in the early grades; a wicked problem; a proposed common measurement; trial outcomes that are not comparable	Henderson and Fantuzzo (2023); Childs and Lofton (2021); Kearney et al. (2026); Graham et al. (2026)

Narrative synthesis, not a causal effect of poverty. Sosu et al. (2021) bring together 55 articles from 1998 to 2019. In high-income countries, lower socioeconomic status goes with more absenteeism, with small effects; the direction also appears in at-risk populations and in low- and middle-income countries. There is more evidence on the family than on the school, and on financial resources — free or reduced-price meals — than on parental schooling. Evidence that the gap changes by reason for absence is limited, and the mediating pathways are little studied. Authors’ inference: closing achievement gaps requires attending to attendance gaps, with an intersectional lens, including after COVID-19.

Razeto Pavez (2026) interviews 55 professionals in Chilean public schools — school leaders, teachers and staff in the social area — and describes four vulnerabilities that operate together: families and lack of support; the community of the school or of the home; transitions and other stressors; and the school as a relational unit. The abstract announces recommendations that it does not detail. Childs and Lofton (2021) state the argument: a reform of teaching alone does not reach the student who is already absent. Chhikara (2025), with ECLS-K:2011, finds that closeness buffers the lack of ethnoracial match; section 7 specifies this. Steyer et al. (2026), on kindergarten health, absenteeism, achievement and racial and ethnic disparities, is cited by title only.

6. Does the reason for absence matter? Excused, unexcused and Mondays

Three empirical studies appear to disagree. They do not ask the same question, they do not look at the same stage and they do not share a design. Choosing a winner would manufacture a controversy that the corpus does not sustain.

Empirical finding. Klein et al. (2022), with the Scottish Longitudinal Study (n = 4419), look at high-stakes examinations at the end of compulsory schooling and of post-compulsory schooling. Total absences are associated with lower achievement at both stages. So too, separately, are unexcused absence, illness, exceptional domestic circumstances and family holidays. First-difference regressions confirm those associations, except for holidays. The authors point to behavioural, health and psychosocial pathways, in addition to lost class time. The abstract gives the direction, not a coefficient by reason. It cannot be said that every absence survives the more demanding model.

A different question. Henderson and Fantuzzo (2023), with ROC curves in a large urban district, find that only unexcused absences classify future risk in the early grades; excused absences do not contribute and dilute the mixture. Gottfried et al. (2025), in primary school, find that any absence is associated with lower achievement and that Mondays are particularly negative, only among excused absences: they read this as not being at school, not as disaffection. The pattern changes across groups of students; the abstract does not say across which.

Pérez-Marco et al. (2025) show that intervention already separates origins: 78 programmes for ages 6 to 18, 46 on school refusal and 32 on unexcused absenteeism or truancy. They define school attendance problems (SAPs) as persistent difficulties in attending or remaining, including anxiety, refusal or fear. Two groups of programmes fit poorly with a single 10 % register. Effectiveness is discussed in section 8.

A reading inference, not a new datum. Klein et al. (2022) estimate achievement gradients; Henderson and Fantuzzo (2023), diagnostic utility; Gottfried et al. (2025), the day of the week within the excused absence. An absence can be associated with worse achievement and, at the same time, worsen a classification if it is added to another type. Sosu et al. (2021) find limited evidence that the socioeconomic gap changes with the reason. The policy inference is to disaggregate type and day before classifying, not to declare a winner.

7. Early childhood and the early grades: where the trajectory begins

The indicator is often discussed as a matter for secondary school. Henderson and Fantuzzo (2023) place the test in the early grades: there, unexcused absences inform risk and mixing them worsens the classification. Chhikara (2025), with ECLS-K:2011, reports significant effects of teacher–student ethnoracial matching on the probability of chronic absenteeism. Closeness and conflict vary by the ethnic and racial group the teacher reports, and greater closeness buffers the adverse effect of the lack of match. There is no coefficient in the abstract. Professional development and diversity training are a recommendation of the authors, not a trial of whether that training reduces absences.

Wu and Weiland (2026) follow 6698 students in Boston, from preschool to third grade. They do not reduce absenteeism: they predict. Extreme gradient boosting (XGBoost) with synthetic minority oversampling (SMOTE) gained 54 percentage points of recall over the logistic regression of the systems in use, and outperformed other algorithms. Without demographic data, precision was comparable. Inference: using those data all the same can label a student, and the abstract does not show that detecting better has reduced absences.

Strelitz et al. (2026) do not separate preschool, but the effect of parental engagement is small and more visible at younger ages. Two works are present by title only: Wei et al. (2023), on absenteeism from preschool to kindergarten among children who receive child-care subsidies, and Steyer et al. (2026), on kindergarten health, absenteeism, achievement and racial and ethnic disparities. It cannot be said whether they confirm or qualify Wu and Weiland (2026), Chhikara (2025) or Henderson and Fantuzzo (2023).

8. What works: what trials and meta-analyses show

There is no single answer on effectiveness. Table 2 states the design, the focus and the direction that each source supports. Where there is no abstract, the row says so. What follows interprets, without a new average.

Source	Design	Intervention or focus	Outcome	Direction and certainty according to the source
Strelitz et al. (2026)	Meta-analysis of trials; GRADE	Mentoring	Days of attendance	0.63 days (−0.34 to 1.61); I² = 64.0 %; low certainty. The interval includes zero.
Strelitz et al. (2026)	Meta-analysis of trials; GRADE	Parental engagement	Days of attendance	0.43 days (0.10 to 0.76); I² = 91.9 %; moderate certainty. A small effect, above all at younger ages.
Strelitz et al. (2026)	Meta-analysis of trials	Targeted, behavioural and socio-emotional approaches	Attendance	Smaller and more heterogeneous bases, with smaller effects.
Graham et al. (2026)	Review of trials; RoB 2; no meta-analysis	Interventions that are mostly multicomponent	Absenteeism, dropout or both	A meta-analysis was not possible. More than half at high risk of bias. No superiority among types.
Pérez-Marco et al. (2025)	Systematic review	78 programmes (46 on school refusal and 32 on truancy), ages 6 to 18	Reduction of the attendance problem	29 showed some level of effectiveness. The abstract does not name the levels.
Jakobsen et al. (2025)	Systematic review and meta-analysis, according to the title	Cognitive-behavioural interventions for school attendance problems	Not reported in the available abstract	Not reported in the available abstract.
Mac Iver et al. (2022)	Regression discontinuity; threshold 0.90	Letters to families (“nudge”)	Change in attendance	Overall: 0.09 (p = .20), not significant. Middle school: 0.34 (p = .044). Full year: not significant.
Grant et al. (2025)	School-level trial (33; 17 and 16)	Restorative practices added to Diplomas Now	Discipline, suspension and chronic absenteeism	Severity: −0.13. Suspension of three days or more: odds ratio = 0.66. Chronic absenteeism only marginal.
Singer (2025)	Daily data; an ecological perspective	Out-of-school suspension	Subsequent attendance	A modest and persistent negative effect, larger if there was no chronic absenteeism and if the disciplinary climate was harsher.
Kjolhede et al. (2025)	Cross-sectional; association, not causality	Rural school-based health centre	Not being at risk of chronic absenteeism	Odds ratio = 1.12 (1.08–1.16). A stronger association in primary school.
Komisarow (2026)	Not reported in the available abstract	School support staff and chronic absenteeism in North Carolina	Not reported in the available abstract	Not reported in the available abstract.

Meta-analytic finding. Strelitz et al. (2026) update the 2022 review of the Education Endowment Foundation (PROSPERO CRD42024610037): 9366 titles, 61 articles from 2000 to 2024 (43 recent and 18 previous) and 57 trials of 54 interventions. Two reviewers placed bias between moderate and low in most; this is not the judgement of Graham et al. (2026). Mentoring: 0.63 days (−0.34 to 1.61), I² = 64.0 %, low certainty, an interval that includes zero. Parental engagement: 0.43 days (0.10 to 0.76), I² = 91.9 %, moderate certainty, above all at younger ages. Interpretation: 0.43 days is a small and highly heterogeneous effect. Most of the trials are from the United States. Inference: mentoring is not preferred for a higher point estimate, because its interval is compatible with a null.

Graham et al. (2026) follow PRISMA 2020, search in February 2024 in five databases and retain 29 trials, 153 740 participants, nine countries and participants under 18. More than half are at high risk of bias; without consistent effects or variances there is no meta-analysis and no superiority among types. They call for clear definitions and for CONSORT. Pérez-Marco et al. (2025) find that 29 of 78 programmes showed some level of effectiveness and shared features of implementation that the abstract does not list; that risk of bias is not transferred to them. Jakobsen et al. (2025) are cited only as a systematic review and meta-analysis of cognitive-behavioural interventions for school attendance problems.

Mac Iver et al. (2022) estimate letters to families by regression discontinuity (threshold 0.90; equivalence between 0.88 and 0.92). The overall effect is 0.09 (p = .20) and the full-year effect is not significant; in middle school it is 0.34 (p = .044), important for the authors according to the What Works Clearinghouse. Design inference: the subgroup does not authorise a general policy. Grant et al. (2025) randomise 33 secondary schools (17 and 16) in 8 districts, in 12 blocks, in 2014–2015 and 2015–2016. The package of restorative practices and Diplomas Now reduced severity (effect size = −0.13) and the odds ratio of suspension of three days or more (0.66). Chronic absenteeism fell only marginally: the package is estimated, not each piece.

9. The school as cause and as response: discipline, health and relationships

The school records the absence and can also produce it. Singer (2025) studies out-of-school suspension in an urban district of the Midwest, with high chronic absenteeism, daily data and an ecological perspective. Finding reported as an effect: suspension has a modest and persistent negative effect on subsequent attendance, larger among those who were not chronically absent and in harsher disciplinary climates. Reducing suspensions can improve attendance and is not enough when social and economic inequalities weigh heavily. Suspending a student for being absent risks the attendance of someone who was still coming; ceasing to suspend does not disarm the rest of the map.

Grant et al. (2025) move discipline more clearly than chronic absenteeism. Together with Singer (2025), the result is not that changing the punishment solves attendance: suspension, used as a corrective, runs in the opposite direction.

Association, not an effect. Kjolhede et al. (2025), in rural northern New York (2015–2016 to 2018–2019), bring together 52 schools, 32 districts and 66 303 students (49.4 % girls): 30 046 with a school-based health centre and 36 257 without one. Except for one cycle, without a centre there was more risk or chronic absenteeism. Adjusting for grade, sex, cycle, economic disadvantage, community wealth and district size, the odds ratio of not being at risk was 1.12 (95 % confidence interval: 1.08–1.16), higher in primary school and where there is more student poverty and more community wealth. Komisarow (2026) studies school support staff and chronic absenteeism in North Carolina: without an abstract, there is no result to cite.

Chhikara (2025) shows that closeness buffers the lack of ethnoracial match. Razeto Pavez (2026) describes, from Chile, the school as a vulnerable relational unit, together with family, community and transitions. Inference: discipline, care and relationship do not substitute for one another. The punitive lever is the one that here has the clearest sign against subsequent attendance.

10. Early warning systems and multi-tiered supports

Wu and Weiland (2026) improve detection, not the attendance that follows an alert. Policy inference: labelling a child as a future chronically absent student stigmatises if the label replaces support, and the more so if the model uses demographic data it does not need in order to predict. Henderson and Fantuzzo (2023) show that, in the early grades, only unexcused absences diagnose and excused absences dilute. Triggering the alert with the aggregated 10 % repeats the assumption that the ROC curve does not support. Not alerting on that sum is an inference tied to the finding, and it is marked as such.

Kearney and Graczyk (2022) do not evaluate an impact: they write a perspective. School attendance problems are complex, heterogeneous and opaque, and multi-tiered systems of support (MTSS) remain nascent. Where chronic absenteeism is very high, they argue, one has to unlearn how the tiers are built and how they are oriented. There are recommendations by tier; the abstract does not list them and they are not reconstructed here. Conceptual framework: the usual pyramid may not describe a massive prevalence. Kearney et al. (2026) move the problem to the measure that would feed any tier or any alert. Without comparability, and without a distinction among engagement, achievement and functional impairment, the assigned tier is as fragile as the 10 %.

Overall inference. Early warning systems and tiers have no mean effect here on days of class. They depend on not mixing absences of different diagnostic value, on not over-labelling and on rebuilding the tiers if prevalence is very high. None of those conditions is settled by a trial in this corpus.

11. Latin America and Mexico: what the corpus allows us to say

The scarcity is a result of the screening, not a formula of modesty. The only Latin American study with an abstract is Razeto Pavez (2026): 55 professionals in public schools in Chile, four vulnerabilities that operate together and a school understood as a relational unit. There are no rates and no trial. Carrying those dimensions to Mexico as if they were a prevalence is a design error; they serve as a neighbouring hypothesis. Mexico appears once and without results. The study by Kozhaya and Martínez Flores (2022) addresses school attendance and child labour in the Full-Time Schools programme (Programa Escuelas de Tiempo Completo). The title fixes the object and authorises neither sign, nor magnitude, nor conclusion.

Sosu et al. (2021) do not give a Mexican datum: the small association between socioeconomic status and absenteeism is confirmed to a good extent in low- and middle-income countries. To ignore stratification would be imprudent; to cite a national figure, impossible. Strelitz et al. (2026) warn that most of their trials are from the United States and that evaluating the interventions in other countries will inform effectiveness. Graham et al. (2026) cover nine countries, without listing them. Nine countries are not a Latin American base. From that perimeter come five policy inferences. They are neither data nor norms.

1. Do not use the 10 % as a sufficient diagnosis. In the early grades, Henderson and Fantuzzo (2023) show that excused absences dilute unexcused ones. This is an inference about the indicator, not a Mexican statistic.
2. Do not import the magnitude of the United States trials as a local promise. External validity is open, and the only Chilean study with an abstract is qualitative.
3. If the bet is on parental engagement, the expectation consistent with Strelitz et al. (2026) is a small, heterogeneous effect, more visible at younger ages.
4. Full-Time Schools remains a question about attendance and child labour, not a known answer. To say that the extended school day raises or lowers attendance goes beyond the corpus.
5. A diagnostic, ecological and non-punitive orientation is consistent with suspension, with relational vulnerability and with the socioeconomic association. Mexico does not have, among these 24 sources, a trial that has tested it. The inference is one of prudence and of an agenda, not of demonstrated effectiveness.

12. Criteria for an evidence-based attendance policy

These criteria are a policy inference. They do not form a programme, they have no effect of their own and they do not replace local evaluation.

1. Disaggregate by type of absence and by day. In the early grades, unexcused absences diagnose and excused absences can dilute them; within excused absences, Mondays are more negatively associated with achievement.
2. Diagnose the level before intervening: health, resources in the home, discipline, relationship and community.
3. If parental engagement is given priority, expect a mean effect of 0.43 days, highly heterogeneous, of moderate certainty and more visible at younger ages.
4. Do not use out-of-school suspension for absences. Lowering suspensions can improve attendance and, on its own, does not close social gaps.
5. Integrate school health as an association — stronger in rural primary school — not as a causal effect ready to be copied.
6. Attend to teacher–student closeness where there is no ethnoracial match. Teacher preparation is a recommendation, not an effect already measured.
7. Alert without labelling. In Boston the prediction held without demographic data, and in the early grades it is unwise to add excused and unexcused absences together.
8. Require trials with CONSORT, comparable outcomes and variances. Without that, there is no cumulative synthesis (Graham et al., 2026).
9. If chronic absenteeism is very high, do not copy the usual pyramid: Kearney and Graczyk (2022) ask that it be unlearned, and there is no trial of that system here. Letters do not scale from a null overall effect either.

13. Limitations

The base is Crossref, with a lexical filter in English and a supplementary search that added only one study. Texts from other databases, or in Spanish, that were not retrieved may be missing. The heuristic that excluded 41 journals of low editorial assurance may have left valid work out. The passage from 317 eligible records to 24 was made without double coding: the judgement is explicit and is not mechanical at every boundary. Jakobsen et al. (2025), Komisarow (2026), Wei et al. (2023), Steyer et al. (2026) and Kozhaya and Martínez Flores (2022) have no usable abstract, and this review does not pretend to know them.

The United States predominates. Evans et al. (2024), Kjolhede et al. (2025) and the literature that Sosu et al. (2021) synthesise are cross-sectional or correlational: they do not identify causal effects even when they adjust covariates or use fixed effects. There is no meta-analysis of our own, also because heterogeneity of definitions already prevented the trials from being pooled in Graham et al. (2026). A publication bias is possible and is not quantified here. Only one Latin American study supplies an abstract, and it is qualitative and Chilean; the Mexican work on the school day remains in the title. That limits what can be said of the region, not the importance of the problem.

14. Conclusions

Chronic absenteeism took centre stage through a simple cut-off, when recovery after the pandemic made attendance urgent. Dee (2024) and Evans et al. (2024) document an increase of 13.5 percentage points between 2018–2019 and 2021–2022. That jump does not sharpen the threshold: excused absences do not diagnose as unexcused absences do (Henderson & Fantuzzo, 2023), measurement remains poorly comparable (Kearney et al., 2026), the indicator can mask a wicked problem (Childs & Lofton, 2021) and the trials do not share an outcome (Graham et al., 2026).

The causes that are sustained are ecological and, often, of modest association. Socioeconomic status is linked to more absences, with small effects and pathways that are little studied (Sosu et al., 2021). In Chile, four vulnerabilities operate together (Razeto Pavez, 2026). The reason matters without cancelling the studies: an achievement gradient in Scotland, except for holidays in first differences (Klein et al., 2022); precision only for unexcused absences (Henderson & Fantuzzo, 2023); negative Mondays within excused absences (Gottfried et al., 2025).

The effects are small and uneven. Parental engagement adds 0.43 days, and mentoring is not distinguishable from a null (Strelitz et al., 2026). The letters have neither an overall effect nor a full-year effect (Mac Iver et al., 2022). The restorative package moves discipline more than chronic absenteeism (Grant et al., 2025), and suspension leaves a negative trace (Singer, 2025). The rural health centre is associated with less risk, without being an effect (Kjolhede et al., 2025). Closeness buffers the lack of ethnoracial match (Chhikara, 2025). Improving recall is not the same as having reduced absences (Wu & Weiland, 2026).

Latin America remains, in this corpus, a Chilean qualitative study, a Mexican title on the extended school day and on child labour, a socioeconomic association in low- and middle-income countries, and trials concentrated in the United States. The inference, not an evaluated programme, is diagnostic, ecological and non-punitive. For as long as the threshold adds absences of different kinds, the number will be clear and the problem will not.

NEXTECH.IA Editorial Lab / Ingeniero Mitre

References

1. Chhikara, D. (2025). The Effects of Student–Teacher Ethnoracial Matching on Chronic Absenteeism: Exploring the Role of Relational Dynamics in Early Education. Early Education and Development, 36(5), 1066–1091. https://doi.org/10.1080/10409289.2025.2484852
2. Childs, J., & Lofton, R. (2021). Masking Attendance: How Education Policy Distracts from the Wicked Problem(s) of Chronic Absenteeism. Educational Policy, 35(2), 213–234. https://doi.org/10.1177/0895904820986771
3. Dee, T. S. (2024). Higher chronic absenteeism threatens academic recovery from the COVID-19 pandemic. Proceedings of the National Academy of Sciences, 121(3), Article e2312249121. https://doi.org/10.1073/pnas.2312249121
4. Evans, W. N., Muchnick, K., & Rosenlund, O. (2024). Virtual Learning in Kindergarten Through Grade 12 During the COVID-19 Pandemic and Chronic Absenteeism. JAMA Network Open, 7(8), e2429569. https://doi.org/10.1001/jamanetworkopen.2024.29569
5. Gottfried, M. A., Woods, S. C., & Peters, S. K. (2025). Linking Student Absenteeism on Different Weekdays to Student Achievement. AERA Open, 11. https://doi.org/10.1177/23328584251400270
6. Graham, P. L., Carrick, A., Littlefair, D., Melo, D., Mulholland, K., Graham, P., Bradley, G., Davies, C., Speyer, R., & Cordier, R. (2026). The structure and methodological quality of randomised controlled trials addressing school absenteeism and dropout in children and adolescents: a systematic review. Educational Research Review, 51, 100791. https://doi.org/10.1016/j.edurev.2026.100791
7. Grant, A. A., Mac Iver, D. J., Byrnes, V., Clark, E., Balfanz, R., & Lofton, R. (2025). Combining Restorative Practices with Diplomas Now: Impacts on Practices, Problems, Suspensions, and Chronic Absenteeism. Journal of Education for Students Placed at Risk (JESPAR), 30(1), 49–73. https://doi.org/10.1080/10824669.2023.2278047
8. Henderson, C. M., & Fantuzzo, J. W. (2023). Challenging the Core Assumption of Chronic Absenteeism: Are Excused and Unexcused Absences Equally Useful in Determining Academic Risk Status? Journal of Education for Students Placed at Risk (JESPAR), 28(3), 259–293. https://doi.org/10.1080/10824669.2022.2065636
9. Jakobsen, S., Tølbøll, K. B., Thastum, M., & Lomholt, J. J. (2025). Cognitive Behavioral Interventions for School Attendance Problems: A Systematic Review and Meta-analysis. Child Psychiatry & Human Development. https://doi.org/10.1007/s10578-025-01847-x
10. Kearney, C. A., Benoit, L., Gonzálvez, C., Heyne, D., Melvin, G. A., O'Toole, C., Genitty, C. G., & Keppens, G. (2026). School attendance and absenteeism: Towards next‐generation common measurement. Review of Education, 14(1), Article e70146. https://doi.org/10.1002/rev3.70146
11. Kearney, C. A., & Graczyk, P. A. (2022). Multi-tiered systems of support for school attendance and its problems: An unlearning perspective for areas of high chronic absenteeism. Frontiers in Education, 7, Article 1020150. https://doi.org/10.3389/feduc.2022.1020150
12. Kjolhede, C., Brunner, W. M., & Sipple, J. W. (2025). School-Based Health Centers and School Attendance in Rural Areas. JAMA Network Open, 8(5), e2510083. https://doi.org/10.1001/jamanetworkopen.2025.10083
13. Klein, M., Sosu, E. M., & Dare, S. (2022). School Absenteeism and Academic Achievement: Does the Reason for Absence Matter? AERA Open, 8. https://doi.org/10.1177/23328584211071115
14. Komisarow, S. (2026). School-based support staff and chronic absenteeism: Evidence from North Carolina. Economics of Education Review, 115, 102819. https://doi.org/10.1016/j.econedurev.2026.102819
15. Kozhaya, M., & Martínez Flores, F. (2022). School attendance and child labor: Evidence from Mexico’s Full-Time School program. Economics of Education Review, 90, 102294. https://doi.org/10.1016/j.econedurev.2022.102294
16. Mac Iver, M. A., Wills, K., Cruz, A., & Mac Iver, D. J. (2022). The Impact of Nudge Letters on Improving Attendance in an Urban District. Education and Urban Society, 54(2), 164–185. https://doi.org/10.1177/00131245211004561
17. Pérez-Marco, M., Gonzálvez, C., Fuster, A., & Vicent, M. (2025). A systematic review of intervention programs for school attendance problems. Children and Youth Services Review, 169, 108091. https://doi.org/10.1016/j.childyouth.2024.108091
18. Razeto Pavez, A. C. (2026). Ausentismo escolar y vulnerabilidad social. Perfiles Educativos, 48. https://doi.org/10.22201/iisue.24486167e.2026.48.61975
19. Singer, J. (2025). The Effect of Suspensions on Student Attendance in a High-Absenteeism Urban District. Urban Education, 60(2), 628–659. https://doi.org/10.1177/00420859231192084
20. Sosu, E. M., Dare, S., Goodfellow, C., & Klein, M. (2021). Socioeconomic status and school absenteeism: A systematic review and narrative synthesis. Review of Education, 9(3), Article e3291. https://doi.org/10.1002/rev3.3291
21. Steyer, L., Omar, J., Sulik, M. J., & Obradović, J. (2026). Kindergarten Health and Absenteeism: Implications for Academic Achievement and Racial and Ethnic Disparities. Early Childhood Research Quarterly, 76, 62–73. https://doi.org/10.1016/j.ecresq.2026.01.011
22. Strelitz, J., Barrett, J. R., Elango, N., Meligy, N., Watson, F., Westbrook, J., Clarke, K., Huw, D., John, O., & John, A. (2026). Interventions to improve school attendance: a systematic review and meta-analysis of evidence from randomised controlled trials. Journal of Epidemiology and Community Health. https://doi.org/10.1136/jech-2025-225671
23. Wei, W. S., Ha, Y., Schneider, K. G., & Joshi, P. (2023). Children's absenteeism from pre-K to kindergarten: A focus on children receiving child care subsidies. Early Childhood Research Quarterly, 64, 368–380. https://doi.org/10.1016/j.ecresq.2023.04.010
24. Wu, T., & Weiland, C. (2026). Leveraging Modern Machine Learning to Improve Early Warning Systems and Reduce Chronic Absenteeism in Early Childhood. Educational Evaluation and Policy Analysis. https://doi.org/10.3102/01623737261419046