The Evidence on Teaching

Phonics Training for English-Speaking Poor Readers (Cochrane Review)

McArthur, G., Sheehan, Y., Badcock, N. A., et al. · 2018

grade Bmeta-analysisindependentreplicatednumbers spot-checked
Sample
14 RCTs/quasi-randomized trials / 923 poor readers
Population
923 English-speaking poor readers (>=1 SD/grade/year below age; no known comorbidity); children/adolescents/adults; AUS/CAN/UK/USA.
Design
Cochrane meta of 14 RCTs; comparator is NO/non-literacy training (e.g. maths), NOT whole language. Random-effects; GRADE-assessed. All outcomes measured immediately post-training (no maintenance data).
Key findings
In poor readers, phonics training reliably improves decoding (nonword 0.67, word 0.51, irregular 0.84, fluency 0.45) but effects on comprehension (0.28) and spelling (0.47) are NOT significant. Note the largest effect is irregular-word reading (0.84), so the honest outcome class is 'word-level reading', not strictly GPC decoding. GRADE low-moderate; immediate post-training only.
Genetic confound
RCT/quasi-random control addresses passive rGE. Decoding-teachable vs comprehension-immovable split mirrors domain-skill vs g/language distinction.
Replication notes
Update of McArthur 2012; conclusions unchanged. Decoding>comprehension pattern consistent with Ehri, Torgerson.
DOI / URL
10.1002/14651858.CD009115.pub3

Effects

OutcomeMetricValueMeasureTimingVsHorizonClass
Non-word reading accuracySMD0.67 (0.26-1.07); GRADE Lowmixedend-of-treatmentbusiness-as-usualend-of-treatmentdomain-skill
Mixed/regular word reading accuracySMD0.51 (0.13-0.90); GRADE Lowmixedend-of-treatmentbusiness-as-usualend-of-treatmentdomain-skill
Irregular word reading accuracySMD0.84 (0.30-1.39); GRADE Moderatemixedend-of-treatmentbusiness-as-usualend-of-treatmentdomain-skill
Word reading fluencySMD0.45 (0.19-0.72)mixedend-of-treatmentbusiness-as-usualend-of-treatmentdomain-skill
Reading comprehensionSMD0.28 (-0.07 to 0.62); NOT significantmixedend-of-treatmentbusiness-as-usualend-of-treatmentdomain-skill
SpellingSMD0.47 (-0.07 to 1.01); NOT significantmixedend-of-treatmentbusiness-as-usualend-of-treatmentdomain-skill

Cited by