Better self-explaining backwards or forwards? Prompting self-explanation in video-based modelling examples for learning a diagnostic strategy

Abstract

Self-explanation prompts in example-based learning are usually directed backwards: Learners are required to self-explain problem-solving steps just presented (retrospective prompts). However, it might also help to self-explain upcoming steps (anticipatory prompts). The effects of the prompt type may differ for learners with various expertise levels, with anticipatory prompts being better for learners with more expertise. In an experiment, we employed extensive modelling examples and different types of self-explanations prompts to teach 78 automotive apprentices a complex and job-relevant problem-solving strategy, namely the diagnosis of car malfunctions. We tested the effects of these modelling examples and self-explanation prompts on problem-solving strategy knowledge and skill, self-efficacy, and cognitive load while learning. In two conditions, the apprentices learned with modelling examples and received either retrospective or anticipatory prompts. The third condition was a control condition receiving no modelling examples, but the respective open problems. In comparison with the control condition, modelling examples did not promote learning. However, we observed differential effects of the self-explanation prompts depending on the learner’s prior knowledge level. Apprentices with higher prior knowledge learned more when learning with anticipatory prompts. Apprentices with less prior knowledge experienced a greater increase in self-efficacy and a higher germane cognitive load when learning with retrospective prompts. These findings suggest using different self-explanation prompts for learners possessing varying levels of expertise.

Differential Development of Professional Knowledge and Problem-Solving Skills During VET: The Role of Initial Cognitive Resources, School-Leaving Certificates, And Sociodemographic Background

“…To promote learning in this domain and to provide apprentices' with these highly relevant skills, additional pedagogical measures could be implemented in both, (inter-)company training and vocational schools. Appropriate simulation-based interventions have been developed and evaluated with substantial effects (e.g., Meier et al, 2022Meier et al, , 2023.…”

Video‐based modeling examples and comparative self‐explanation prompts for teaching a complex problem‐solving strategy

“…These freed cognitive capacities can then be used for learning (i.e., for GCL), which explains the frequently found beneficial effect on learning outcomes (Renkl, 2014;Sweller, 2006). However, to benefit from the reduced ECL, learners must actively use these freed-up capacities for learning (GCL), that is, they must engage in generative Frerejean et al, 2018) and other studies finding no beneficial effects (e.g., Meier et al, 2023). Thus, further research in this regard is necessary to further investigate the boundary conditions of examplebased learning (Sweller, 2023).…”

“…Thus, a filled-in diagnosis plan represents a summary of a diagnostic process. We developed modelling examples in a screencast format showing an expert applying the diagnostic strategy and filling out a diagnosis plan in a computer simulation (Gschwendtner et al, 2009;Meier et al, 2022Meier et al, , 2023. While studying the modelling examples, apprentices received selfexplanation prompts that referred to the filled-in diagnosis plans that served as static representations of the problem-solving process.…”

“…diagnoses of automotive malfunctions. For this test, we used a selection of five out of 24 items from the diagnosis-relevant reception competence (DRC) test by Norwig et al (2021), as in our previous study (Meier et al, 2022(Meier et al, , 2023. To prevent floor and ceiling effects, we selected items for their midrange solution range (ranging from 32% to 71% in Norwig et al, 2021) and with the highest item-total correlation (>0.43 for all 5 items in Norwig et al, 2021).…”

“…Besides these interaction effects, we found three main effects of the timepoint on dependent variables that were not affected by interaction effects, namely on the participants' strategy application test score (large effect), their interest in diagnoses (medium effect), and their perception of diagnoses as a challenge (medium to large effect). In our previous study (Meier et al, 2022(Meier et al, , 2023 we assessed the apprentices' motivation (i.e., self-efficacy, interest, perception of challenge, and incompetence fear). In the present study, we assessed the apprentices' motivation with the same items to ensure that neither the modeling examples nor the prompts had negative effects on the apprentices' motivation.…”