The relation between parents’ involvement in children’s schooling and children’s adjustment: A meta-analysis

448 independent studies covering 480,830 families

Children and adolescents across ages, ethnicities and socioeconomic strata, with parents' NATURALLY OCCURRING (not experimentally induced) involvement in schooling.

Psychological Bulletin 145(9). The largest synthesis of parental involvement ever assembled, and deliberately restricted to naturally occurring involvement — i.e. it is a correlational meta-analysis by design and inherits every selection and passive gene-environment confound in the underlying studies. Grade D follows from that, not from execution. It earns its place here for one reason: it disaggregates involvement by TYPE, and one type behaves completely differently from the rest. Longitudinal analyses were run alongside concurrent ones and the positive associations were maintained over time.

Across 448 studies and 480,830 families, parents' naturally occurring involvement in schooling shows small positive correlations with academic adjustment (r = .13 to .23), social (r = .12) and emotional adjustment (r = .17), and a negative one with delinquency (r = -.15). Different types of involvement — attending school events, discussing school with the child — were similarly positively associated with academic adjustment. The single exception is the one that matters for parent-delivered instruction: parents' HOMEWORK ASSISTANCE was NEGATIVELY associated with children's achievement, r = -.15, while remaining slightly positive for engagement (r = .07) and motivation (r = .05). There was little variation by age, ethnicity or SES. Note the ceiling on everything here: an r of .13 to .23 across observational studies of families is exactly the magnitude this archive expects heritable confounding alone to produce, so the positive findings carry almost no causal weight. The homework-help reversal is more informative precisely because the confound runs the WRONG way — struggling children attract more homework help, so the sign is at least partly reverse causation rather than harm.

HIGH throughout, by construction — every included study measures involvement that parents chose. The same parents who attend school events and discuss school supply the alleles. The homework-assistance reversal has an additional and probably dominant confound in the other direction: parents help more when the child is struggling, so r = -.15 is a mixture of selection into help and any true effect of help, and the meta-analysis cannot separate them.

The homework-help negative replicates across independent syntheses and data sources: Fernandez-Alonso et al. 2022 finds d = 0.23 favouring students who receive LESS family homework help across four PISA cycles (2009-2018) and 180 effects; Xu et al. 2024's three-level meta reports r = -0.064 overall with autonomy-supportive help positive and content/control help not. Maloney et al. 2015 supplies the mechanism candidate on the maths side. Every one of these is observational and shares the reverse-causation problem, so the replication establishes the ROBUSTNESS of the correlation, not its interpretation.