The Efficacy of Student-Centered Instruction in Supporting Science Learning

140 teachers randomised (70/70); 125 teachers analysed (66 intervention, 59 comparison); 2,594 students (1,418 intervention, 1,176 comparison)

Grades 4-5 (about 40% grade 4), one central Florida county, 2007-08 and 2008-09; 31% free/reduced-price lunch, 62% White, 3% English learners.

Two-step cluster randomisation: volunteer teachers were matched on student demographics and grade, then randomly assigned. The treatment is NOT open discovery — it is the GEMS Space Science Sequence, a fully scripted commercial curriculum of 24 sixty-minute sessions with an explicit focus on models and evidence, supported by four days of pre-service PD, a three-hour refresher, and a mid-unit science coach. The comparison arm used the district text and is described as "a more didactic presentation... including direct instruction, reading of text, and students answering very focused questions" on the same content. So this is a scripted, heavily-scaffolded guided-inquiry curriculum versus a textbook curriculum, not student-led inquiry versus teaching. Crucially the outcome measure is NOT treatment-aligned: it is the MOSART (Misconceptions-Oriented Standards-Based Assessment Resources for Teachers) Space Science Content test, developed by a third party, aligned to national content standards and reviewed by science faculty. WWC reviewed this trial (as Granger et al. 2010 NARST) and rated it "meets WWC evidence standards without reservations"; teacher attrition 11% overall / 10% differential, student attrition unknown. Numbers below are WWC-recomputed, not the authors' framing.

The published framing in Science ("learning outcomes were higher for students... through a student-centered approach") is correct but the magnitude is small and it does not last. WWC-verified effect on the externally-developed Space Science Content test was d=0.17 (+7 percentile points, p<0.001) immediately after the unit — and d=0.07 (p=0.19, NOT significant) at the five-month follow-up. This is the best-identified randomised science-instruction contrast in this tranche and it is also the one measured on a non-aligned test; the pairing of those two facts is the point. Secondary measures the authors emphasise (Models and Evidence Questionnaire, Views of Scientific Inquiry, science attitudes) did not meet WWC inclusion criteria for a science achievement outcome.

Low. Teachers were randomly assigned within matched blocks and both arms drew from the same district population; student-level attrition is the residual exposure.

No independent replication of the GEMS Space Science Sequence located. WWC's intervention report rests on this one study and rates the extent of evidence "small". The fadeout pattern (0.17 to 0.07 over five months) matches the archive's general fadeout finding rather than anything specific to inquiry.