Performance Expectancies Moderate the Effectiveness of More or Less Generative Activities Over Time

Frontiers in Psychology

article

Marc-André Reinhard¹  Kristin Wenzel¹ 

[1] Department of Psychology, University of Kassel

关键词: desirable difficulties;    problem-solving;    generation effect;    worked-examples;    performance expectancies;    meta-cognition;    long-term learning;   

DOI  :  10.3389/fpsyg.2019.01623

学科分类：社会科学、人文和艺术（综合）

来源: Frontiers

【 摘 要 】

We examined if the benefits of generation for long-term learning depend on individual differences in performance expectancies (PEs) prior to learning. We predicted that a greater generative activity (problem-solving) compared to less generative activity (worked-examples) should be more effective for pupils with higher PEs, especially in the long run. As a comparison group for problem-solving, we implemented a special type of worked-examples that decreased engaging in self-explanations, because our main prediction focused on PEs moderating the long-term effectivity of less versus greater generative activities. We tested students’ immediate and delayed performance (after 3 months) using coherent curricular materials on linear functions in a sample of eighth graders (advanced school track). The results were partly in line with our predictions: Although we found no moderation of PE and generative activity, we obtained the predicted 3-way interaction of PE, generative activity, and time. Immediately, greater generative activity (problem-solving) was beneficial for pupils with higher PEs, while for pupils with lower PEs, problem-solving versus worked-examples did not differ. In the delayed test, this pattern reversed: for lower PEs, greater generative activity outperformed less generative activities, but there was no difference for higher PEs. Unexpectedly, the initial advantage of problem-solving for higher PEs could not be maintained, decreasing over three subsequent months, whereas the performance in the worked-example condition remained at a comparable level for higher PEs. The change in performance in the problem-solving condition for lower PEs was descriptively less pronounced than in the worked-example condition, but statistically not different. We further investigated the effects of problem-solving and worked-examples on changes in PEs after learning and after testing, hinting at gradual decrease in PEs and greater metacognitive accuracy in the problem-solving condition due to a reduction of overconfidence.

【 授权许可】

CC BY