The Effect of Family Literacy Interventions on Children's Acquisition of Reading From Kindergarten to Grade 3: A Meta-Analytic Review

16 intervention studies, 1,340 families

Children from kindergarten to grade 3, with and without reading difficulties, across socioeconomic groups

Meta-analysis of intervention studies only, which is what makes it usable — the family-literacy literature is otherwise almost entirely correlational. The moderator that matters is WHAT the parent was asked to do, and it separates the effect by three levels of instructional specificity. Three cautions travel with the numbers. The cell effects rest on 2-7 small studies each; effect sizes were NOT homogeneous (0.07 to 2.02 across studies) and the authors themselves write that "caution should be used in interpreting the results"; and standardized tests yielded SMALLER effect sizes than researcher-designed tests, which is the archive's 2x measure-type inflation appearing inside the meta-analysis. d=1.15 from a handful of small studies is a red flag under this archive's own benchmarks, and the 2025 Dahl-Leonard meta-analysis of school-aged family-implemented literacy instruction returns g=0.36 from group designs. The precursor National Institute for Literacy report (2006, 14 studies, 1,174 families) gives the same three cell values with an overall weighted d=0.68 (95% CI 0.56-0.81).

Parent involvement has a positive effect on children's reading acquisition, and the effect is entirely a function of instructional specificity. Parents TUTORING children on specific literacy skills: d=1.15 (7 studies). Parents LISTENING to their children read: d=0.51-0.52 (5 studies). Parents READING TO their children (3 studies in the published version, 2 in the precursor report): d=0.18, not different from zero. Overall pooled effect d=0.65 (CI 0.53-0.76) in the published version. Effects did not differ by grade (kindergarten vs grades 1-3), reading level, socioeconomic status, random assignment, sample size, posttest timing or country. Two moderator results cut against the intuitive reading: providing SUPPORTIVE FEEDBACK to parents during the intervention "did not produce better child outcomes" (8 studies with vs 6 without), and less parent training was associated with LARGER effects — though the authors decline to interpret that because four of the six less-training studies were teaching interventions, so training is confounded with intervention type. The ordering across the three activity types is the durable finding: the same parent, the same child and roughly the same time produce a large effect, a moderate effect or nothing depending only on whether the activity is specified as instruction.

Low for the ranking — these are randomized or controlled interventions in which the family is held constant and the activity varies. Higher for the absolute magnitudes, which come from small studies with volunteer families.

The shared-reading null replicates hard: Noble et al. 2019 finds g=0.23 against a passive control but g=0.03 against an active control. The specificity gradient — instruction beats exposure — is the same pattern the archive's phonics and explicit-instruction topics find when a teacher delivers it. The MAGNITUDES do not replicate: Dahl-Leonard et al. 2025, covering family-implemented literacy instruction with school-aged children, returns g=0.36 from group designs against g=1.50 from single-case designs, which is where numbers of this size come from.