Effects of Reflection, Worked Examples, and Reflection Examples on Learning and Transfer of Reasoning and Judgment Tasks

• In Event: The Form and Function of Transfer in Diverse Contexts

Abstract

This presentation will focus on learning and transfer of reasoning and judgment tasks. Many people perform poorly on “classic” reasoning and judgment tasks, because they tend to respond automatically and base their responses on the inappropriate cues. This study investigated whether instructions to reflect could counteract automatic responses and improve performance. For example, in the domain of medical reasoning and judgment, we showed that structured reflection could counteract biased diagnostic reasoning. However, that study was conducted with residents who had the knowledge required to solve the case correctly. It is questionable whether reflection is effective when people lack prior knowledge. In this case, studying worked examples that show and explain the right answer might be more effective for learning.

Yet, ultimately, one would hope the effects of instruction would not be confined to the trained tasks, but would transfer to other task types. Since worked examples only provide students with knowledge of the right answer, transfer is not likely. Hence, we developed “reflection examples” in which the “model” originally provided presented commonly occurring incorrect answers, individuals engaged in answering structured reflection questions, and finally through reflection arrived at the right answer. This was expected to not only enhance learning, but also foster transfer.

Dutch students (72) completed a retest phase, learning phase, and posttest phase. Participants were randomly assigned to one of four learning phase conditions: problem solving (PS), structured reflection (Re), worked examples (WE), and reflection examples (ReE). The pretest and posttest consisted of eight multiple-choice tasks: two conditional syllogisms (reasoning), two Wason selection (reasoning), two conjunction fallacy (judgment) and two base-rate fallacy (judgment). The learning phase used the same tasks as the pretest. Half of the learners received the syllogism and base-rate fallacy tasks again; the other half the Wason selection and conjunction fallacy tasks (to prevent potential effects on learning/transfer to be due to task specific characteristics). Learning was assessed by examining the pretest to posttest performance gains on tasks encountered in the learning phase, and transfer was assessed as performance gains on the tasks not encountered in learning phase.

Results show that worked examples that show and explain the correct answer were most effective for learning. For transfer, however, our hypothesis that reflection examples would be most effective was not confirmed; only on judgment tasks did they seem advantageous and only compared to reflection, although it should be noted that we only used an immediate posttest. Potentially, reflection examples, by starting with an initially incorrect answer, may hamper initial performance. Because they teach students what questions to ask themselves, they might show longer-term benefits on transfer. Hence, we will repeat this study later this year with a delayed posttest.

Authors

• Tamara Van Gog, Erasmus University

• Luh Anjani Kusuma, Erasmus University

• Sofie Loyens, Erasmus University

• Martine Baars, Erasmus University Rotterdam

• Anita Heijltjes, Avans University of Applied Sciences

• Silvia Mamede, Erasmus University