Strengthening Verbal Skills through the Use of Classroom Drama: A Clear Link

200 studies since 1950 identified, 80 selected; seven separate meta-analyses testing nine hypotheses

School-age children in classroom drama interventions where texts are enacted

The drama arm of Harvard Project Zero's REAP project and the single strongest positive finding in the whole arts-transfer literature. Inclusion required a verbal-achievement measure, an experimental design, and enough information to compute an effect size. Grade C. Two honesty flags: the effect sizes below were obtained from the detailed study summary in Deasy's Critical Links (2002, ERIC ED466413) rather than from the original article, which is paywalled, so they are second-hand; and Critical Links reports CONFIDENCE INTERVAL RANGES rather than point estimates, so no single headline r exists in this record.

Six of seven meta-analyses positive, one null. Confidence-interval ranges for the effect (as r): oral story understanding and recall .16 to .34 (17 reports); WRITTEN story understanding .37 to .73; reading achievement on standardized tests .11 to .29; reading readiness .15 to .36; oral language development .20 to .41; writing achievement .09 to .52 (8 reports); and VOCABULARY -.07 to .19, the null. Larger effects appeared for structured plots, low-SES populations and remedial readers - and, tellingly, for NON-EXPERIMENTAL designs and for journal-published studies, so design weakness and publication status both inflate the estimate. Podlozny's own transfer caveat is the line advocates never quote: "transfer of skills from one domain to another is not automatic: it needs to be taught."

Not addressed. Design quality varies across the 80 reports and the meta-analysis itself shows non-experimental designs producing larger effects, which is the signature of uncontrolled selection.

Recorded as `mixed`. The direction reproduces in Lee et al. (2015) and Lee, Enciso and Brown (2020), but the latter shows the standardized-test estimate is d = 0.29-0.35 against d = 0.89 for researcher-made measures - so the upper half of Podlozny's ranges is probably measure artefact. The Critical Links reviewers, applying stricter comparison-group criteria, distilled the same field to 19 studies and FAILED to reproduce the reading-readiness finding. And the two randomised tests with standardised outcomes go the other way: the EEF Speech Bubbles drama trial (n = 821) was null on reading and on oral narrative recall, and Hautala et al. (2023) found an active oral-reading control outperformed readers' theatre.