Effects of instructional support within constructivist learning environments for elementary school students' understanding of "floating and sinking."

161 third-grade students in three groups (high instructional support, low instructional support, uninstructed baseline)

German elementary schools, grade 3 (age ~8-9); topic floating and sinking (density and buoyancy force).

The most decision-relevant study in this tranche because it is on CHILDREN, on a real curriculum unit, with a ONE-YEAR FOLLOW-UP - the combination almost nothing else in the conceptual-change literature offers. Repeated-measures design: pretest, post-test, 1-year follow-up. Both instructed arms received the same constructivist, hands-on curriculum (720 minutes, i.e. 11 periods of 45 minutes, roughly half a semester of science); the arms differed only in SEQUENCING of content and in the teacher's cognitively structuring statements - i.e. this is a guidance-dose contrast inside a constructivist environment, NOT inquiry versus direct instruction. The baseline group received no instruction on the topic, so the instructed-vs-baseline contrast is against nothing rather than against an alternative use of the same 12 hours. Group assignment was at class level and is not described in the record we could read as randomised; the curriculum was developed by the Münster group to which two authors belong, hence developer-involved. Outcome is a researcher-built test of density and buoyancy understanding, scored for both misconception reduction and adoption of scientific explanations. We could not obtain the full text (paywalled, no OA copy) and therefore have no effect sizes - only the reported direction and significance.

At post-test both instructed groups gained significantly on understanding of density and buoyancy force relative to the uninstructed baseline group - so 12 hours of guided hands-on instruction moves 8-year-olds on a concept US curricula avoid at this age. ONE YEAR LATER, the HIGH instructional support group was superior to the LOW instructional support group on two separate outcomes: reduction of misconceptions and adoption of scientific explanations. This is the cleanest age-appropriate evidence in the tranche that (a) conceptual change in children survives a year, and (b) the durable difference is produced by TEACHER STRUCTURING, not by the hands-on activity that both arms shared - the same guidance finding the archive already records from Furtak 2012, Alfieri 2011 and Lazonder & Harmsen 2016, now with a 12-month horizon. The effect at follow-up is a high-vs-low-guidance contrast, so it is an active-alternative comparison, which makes it more impressive than a business-as-usual gap of the same size. Effect magnitudes were not obtainable.

Medium. Class-level allocation without documented randomisation leaves selection between arms uncontrolled, though the uninstructed baseline makes the instruction effect itself hard to explain by heredity.

Not directly replicated. The intervention is included in Schwichow & Zoupidis 2024, in whose corpus it is one of the longest-duration studies (720 minutes), and that meta-analysis' duration moderator (+0.06 per hour) is consistent with its size.