Peer-reviewed article

Seeing the Forest because of the Trees: New Perspectives on Daycare Center Closures, Maternal and Child Well-being and Early Childhood Education Quality through Machine Learning

Franz Stephan Neuberger, Mariana Grgic, Johannes Wieschke, Franz L. Classe, Susanne Kuger · 6 Jun 2025

Summary

Written by AI, not yet AICS's view · claude-haiku-4-5-20251001

This study examines how unexpected daycare closures affect mothers and children in Germany using machine learning analysis of survey data from nearly 15,000 mothers. The research found that daycare closures reduce both maternal well-being and children's well-being, with longer closures causing stronger negative effects. The study shows that these effects work mainly through parents' perception of educational quality rather than through structural changes, and that families without alternative childcare arrangements suffer the most.

Key points

Short daycare closures reduced maternal well-being by 0.097 SD; longer closures showed stronger effects of 0.192 SD
Children's daycare-specific well-being was significantly affected, with larger effects from longer closures (0.178 SD)
About 53% of effects on maternal well-being and up to 100% of effects on child outcomes were mediated through parents' perception of process quality
Families without alternative childcare and those with inflexible work schedules experienced disproportionately worse outcomes
Daycare closures represent a chronic structural problem affecting 44% of German families

Limitations

The study relies on self-reported well-being measures and mother-assessed child well-being rather than independent child assessments. The research cannot establish whether the observed effects persist long-term or whether other unmeasured factors contribute to the outcomes.

Abstract

This paper estimates the causal eﬀects of unexpected daycareclosures on maternal and child well-being using machine learning methodsfor causal inference. Our identification strategy exploits the conditional in-dependence of closure assignment given observed covariates, with closuresprimarily driven by external staﬃng constraints rather than family charac-teristics. Using survey data from 14,867 mothers in Germany, we examineboth maternal self-reported well-being (WHO-5) and mother-assessed childwell-being (KINDL scales). Our causal forest analysis reveals significantdose-response relationships: short closures reduce maternal well-being by-0.097 standard deviations (SD) (95% CI: [-0.136, -0.058]), while longer clo-sures show stronger eﬀects -0.192 SD (95% CI: [-0.261, -0.123]). Children’soutcomes follow similar patterns, with daycare-specific well-being showinglarge eﬀects (-0.178 SD for longer closures). Mediation analysis reveals thatclosure eﬀects operate primarily through perceived process quality ratherthan structural disruptions. For maternal well-being, approximately 53% ofthe eﬀect is mediated through process quality perceptions, while for childoutcomes this proportion reaches up to 100%. Our heterogeneity analysisusing decision trees identifies vulnerable subgroups: families without alterna-tive childcare arrangements and those with inflexible work schedules experi-ence disproportionately worse outcomes. While unexpected closures emergedduring COVID-19, today’s closures represent chronic structural problems af-fecting 44% of German families. These disruptions fundamentally reshapeparental trust in early childhood institutions, highlighting urgent needs forpolicy interventions to preserve educational equity.

Findings the Hub has extracted

Each claim below is labelled by its type of design and links back to this report.

DevelopmentProtection

Problem

Unexpected daycare center closures and their effects on maternal and child well-being

Intervention: Daycare center closures of varying duration
Population: 14,867 mothers and their children in Germany

Key finding

Short closures reduced maternal well-being by 0.097 SD (95% CI: -0.136 to -0.058)

Practitioner