The Evidence on Teaching

A Response to Recent Reanalyses of the National Reading Panel Report: Effects of Systematic Phonics Instruction Are Practically Significant

Stuebing, K. K., Barth, A. E., Cirino, P. T., Francis, D. J., & Fletcher, J. M. · 2008

grade Ccritiqueindependentmixednumbers spot-checked
Sample
40 studies / 225 comparisons (NRP corpus)
Population
Same NRP corpus.
Design
Reconciles NRP vs Camilli via an a-b-c identity: a=systematic-vs-no-phonics, b=some-vs-no-phonics, c=systematic-vs-some (c=a-b).
Key findings
Shows NRP's ~0.41 and Camilli's ~0.24 are the SAME data answering different questions ('vs no phonics' vs 'vs some phonics') and are statistically compatible. Both camps agree teaching GPCs beats teaching none; the live dispute is the marginal value of SYSTEMATIC over merely explicit phonics, which is small and not cleanly demonstrated.
Genetic confound
N/A (experimental corpus).
Replication notes
Directly rebutted by Camilli, Kim & Vargas (2008); the estimand dispute has no agreed single point estimate. Decomposition verified against full text.
DOI / URL
10.1037/0022-0663.100.1.123

Effects

OutcomeMetricValueMeasureTimingVsHorizonClass
Systematic phonics vs no-phonics control (a)d0.514 (Camilli Table 5) / 0.443 (regression)mixedend-of-treatmentbusiness-as-usualend-of-treatmentdomain-skill
Some phonics vs no-phonics (b)d0.243mixedend-of-treatmentbusiness-as-usualend-of-treatmentdomain-skill
Systematic vs some phonics (c = a-b)d0.271 (agrees with Camilli's 0.24)mixedend-of-treatmentactive-alternativeend-of-treatmentdomain-skill
NRP-equivalent average (a+c)/2d0.39 (~ NRP 0.41)mixedend-of-treatmentunclearend-of-treatmentdomain-skill

Cited by