What Works to Keep Students in School? A Meta-Analysis of Interventions to Reduce School Absenteeism
Li D, Singh S, Jeldi N, Chang S, Zhu M, Dailey K, Karabukayeva A, Cai C · 2026
grade Cmeta-analysisindependentreplicated
Sample
13 RCTs and quasi-experimental studies published January 2020 - March 2025
Population
PreK-12 US students
Design
Random-effects meta-analysis of absenteeism outcomes only, with meta-regression over ten moderators grouped by ecological systems theory. Pools RCTs with quasi-experiments, so it inherits the identification quality of the weaker half.
Key findings
Overall Hedges g = 0.091 (95% CI 0.012-0.170) on absenteeism - small, barely excluding zero. Substantial heterogeneity (I2 = 73%). Study setting was the only significant moderator, with stronger effects in single-school implementations, which is the familiar pattern of effects shrinking as implementation scales. Absenteeism only; no achievement outcome pooled.
Genetic confound
Medium - mixes randomised and quasi-experimental designs. Not the reason for exclusion.
DOI / URL
Effects
| Outcome | Metric | Value | Measure | Timing | Vs | Horizon | Class |
|---|---|---|---|---|---|---|---|
| School absenteeism | Hedges g (random effects) | 0.091 (95% CI 0.012-0.170) | standardized | post-intervention | business-as-usual | end-of-treatment | behaviour |
| Heterogeneity | I2 | 73% | standardized | post-intervention | unclear | end-of-treatment | behaviour |
| Moderator - single-school vs multi-site implementation | moderator significance | only significant moderator; larger effects in single-school implementations | standardized | post-intervention | unclear | end-of-treatment | behaviour |