Training for fostering knowledge co-construction from collaborative inference-drawing

Groups typically have difficulties drawing inferences that integrate individuals' unique information (collaborative inferences) and thus yield a true assembly bonus. An experiment with 36 dyads of university-level students in four training conditions showed, particularly in untrained dyads, that collaborative inferences were less likely to be drawn than two other types of inferences that did not require collaborative information processing. Reflected case-based training led to more inferences during subsequent testing. Testing performance was highest if the training had also included explicit strategy instruction in combination with adaptive tutoring. However, explicit strategy instruction without adaptive tutoring offered no advantage. (C) 2010 Elsevier Ltd. All rights reserved.