'Consider the Opposite' - Effects of elaborative feedback and correct answer feedback on reducing confirmation bias - A pre-registered study

Authors

Suzan van Brussel
Miranda Timmermans
Peter Verkoeijen
Fred Paas, University of WollongongFollow

RIS ID

141577

Publication Details

van Brussel, S., Timmermans, M., Verkoeijen, P. & Paas, G. (2020). 'Consider the Opposite' - Effects of elaborative feedback and correct answer feedback on reducing confirmation bias - A pre-registered study. Contemporary Educational Psychology, 60 101844-1-101844-11.

Abstract

Unbiased reasoning is considered an essential critical thinking skill that students need to possess to face the future challenges in their work and life. Confirmation bias, which is the tendency to selectively attend to information that is consistent with held beliefs, presents a significant thread to unbiased reasoning. An effective strategy to reduce confirmation bias is the 'consider-the-opposite'-strategy (COS). The central question of this pre-registered study was whether providing elaborative, worked example feedback after COS practice would lead to a better performance on previously practised and transfer tasks than correct-answer feedback. Participants were 132 university students who took a confirmation bias pre-test, watched an instructional video on COS afterwards and next received either worked example feedback or correct answer feedback on practice tasks, practised only, watched the instruction only or received no treatment. Finally, all participants took a learning test assessing their skill to avoid confirmation bias, and a transfer test assessing whether they could apply this acquired skill to problems containing other biases. Results revealed no differences on the learning test between both feedback conditions, but students who received feedback scored significantly higher on the confirmation bias problems than students who did not receive feedback. We carried out our pre-registered analysis plan, but due to the low reliability of particularly the pre-test, we carried out an additional exploratory analysis on subsets of post-test items and a subset of transfer test items. Results on learning revealed the same pattern as the planned analyses. However, we found no differences between any of the conditions on transfer.