A meta-analysis of national research: Effects of teaching strategies on student achievement in science in the United States

61 US studies published 1980-2004

US K-12 science classrooms; achievement outcomes.

Inclusion required a US setting, an experimental or quasi-experimental design, and reported (or computable) effect sizes. The eight "teaching strategy" categories are the authors' own inductive coding of what each study did, and the authors concede the categories "were not discrete and often overlapped" — so the per-category numbers below are not independent estimates of separable treatments. Their definition of Inquiry Strategies is broad and guidance-agnostic: "Teachers use student-centered instruction that is less step-by-step and teacher-directed than traditional instruction; students answer scientific research questions by analyzing data (e.g., using guided or facilitated inquiry activities, laboratory inquiries)". The magnitudes are the problem. Against business-as-usual classroom controls, d=0.65 for inquiry and d=1.48 for "enhanced context" are far outside anything a well-identified, independently measured classroom trial in science has produced (compare Granger 2012 at d=0.17 on an external test, and the EEF TDTS effectiveness trials at +0.01 and +0.02). METHODOLOGY treats d>0.60 from small studies with treatment-aligned researcher-designed outcomes as a red flag, not a triumph; that is the correct reading here. No funnel-plot or PET-PEESE correction is reported. Publication comes from a Texas A&M group with no stake in any of the pooled programmes.

Eight strategy categories, all reported as significant: Enhanced Context 1.48, Collaborative Learning 0.95, Questioning 0.74, INQUIRY 0.65, Manipulation 0.57, Assessment 0.51, Instructional Technology 0.48, Enhanced Materials 0.29. This is the most-quoted "inquiry beats traditional teaching in science" number, and it should be read as an upper bound produced by aggregating 61 mostly small US studies with predominantly aligned, study-built outcome measures. The rank order is also uninterpretable as a menu: the categories overlap by the authors' own admission, so "questioning beats inquiry" is not a finding.

Low. Within-study instructional contrasts; the operative confounds are small-study bias and outcome-measure alignment, not heredity.

Minner 2010 cites this meta as directionally concordant with its own (much weaker) vote-count result. Furtak 2012, restricted to experimental and quasi-experimental studies, reports a smaller inquiry effect (d=0.50) on similarly aligned measures, and resolves it into a guidance contrast. No independent re-analysis of Schroeder's dataset was located.