Scientific reasoning vs multivariable causal inference (2011)
Bachelor Thesis Boris Ney
This study aims to replicate the finding that isolating the effects of multiple independent variables step-by-step using the control-of-variables strategy (COV) does not lead to understanding-of-multivariable-causality (UMC). The difference in learning gain between a scientific teaching instruction following a multivariable-causal-inference (MCI) approach and a scientific-reasoning (SR) approach was assessed. Additional posttest items measuring understanding of the multivariable structure of the task (UMST) were used to compare in how far both conditions were able to identify causal and non-causal independent variables. In a pre-posttest design twenty-three children were randomly assigned to one of the two conditions. Both groups of learners progressed significantly, but a significant difference in learning gains between the two conditions was found, in favour of learners in the SR condition. Scores on the additional posttest items nevertheless indicated that the learners had poor insights into the multivariable structure of the task with no differences between both conditions. In addition, COV-experts from both conditions had no above-average scores on these posttest items, indicating poor UMST. Results are discussed and limitations of the current study as well as alternative explanations for its findings are outlined, closing the article with suggestions for future research in this area.
First supervisor: Pascal Wilhelm
Second: Ard Lazonder
2011 - University of Twente