Producing Scientific Tests: What Develops and When?.

Developmental changes in the understanding and use of the logic of experimental control were addressed with three tasks in a longitudinal study. Children in Grades 2 to 5 produced experimental tests and chose/evaluated tests made by hypothetical others. Cress-task and longitudinal analyses suggested that whereas the understanding of experimental control increased substantially between Grades 2 and 4, using this understanding to produce controlled tests lagged behind. On the basis of an analysis of some of the sources of individual differences, a training condition that facilitated the active representation and mental combination of variable dimensions was tested. Substantiai test production improvement aftei training suggests that children's failure to produce controlled experiments arises from their failure to actively represent the problem space, rather than from a fundamental lack of understanding of the logic of experimental control. "PERMISSION TO REPRODUCE THIS MATERIAL HAS BEEN GRANTED BY Merry Bullock TO THE EDUCATIONAL RESOURCES INFORMATION CENTER (ERIC)" BEST COPY AVAILABLE U.S DEPARTMENT OF EDUCATION Office 01 Educattonal Research and Improvement EDUCATIONAL RESOURCES INFORMATION CENTER (ERIC) Thts document has been reproduced at rece.vid horn the vemon or organtiatIon rutcpnahng II I) Mtn°, changes have been made to improve reproduction Quallty Points of new Or opinions staled tn thus docu. ment do nal necessarily represent h Kuhn et al., 1988). Yet, younger grade schoolers seem to understand some of the basic concepts of.experimentation. For example, they understand that one must vary the presence or abseraie of a potential cause and can recognize when variables are not controlled (e.g., Bullock, 1991), and they can produce a critical test to distinguish between two potential causes when no other variables are present (e.g., Sodian et al., 1991). There is also controversy over the reasons for children's failures to produce controlled tests: because they lack a conceptual understanding of experimental control, because they misconstrue the task to be one of hypothesis confirmation rather than hypothesis testing, or because they lack general analytic or representational skills. To address these issues, three tasks assessing the understanding of the logic of experimental control were presented in an ongoing longitudinal project on children's scientific reasoning. In all three tasks, understanding of experimental control was assessed by production measures (children were asked to test potential causal relations in a multivariable situation) and by choice/evaluation measures (children were asked to choose or evaluate tests made by hypothetical others). To assess developmental changes, one task (story task) was presented in