The present study assessed the role of context in the acquisition and
transfer of a mathematical strategy. One hundred and six children were
assigned to four conditions: direct strategy instruction, guided disc
overy, direct teaching plus discovery, or a control condition. The int
ervention consisted of fourteen sessions during which the number-famil
y strategy, useful for addition and subtraction, was taught. Third gra
de students in the guided discovery condition performed better than th
ose in the direct instruction condition on far transfer problems that
measured deep conceptual understanding. Students who had total or part
ial exposure to guided discovery held stronger beliefs and adopted mor
e positive goals about the importance of mathematical understanding an
d peer collaboration, attributed less importance to task extrinsic rea
sons for success, and reported greater use of deep processing strategi
es than students exposed to direct, explicit instructions. Finally, st
udents in the discovery conditions were able to communicate more effec
tively during problem solving than students in the direct instructions
condition.