Understanding scientific phenomena requires learners to construct mental models of causal systems. Simulation-based discovery learning offers learners the opportunity to construct mental models and test them against the behavior of a simulation. The purpose of this study was to investigate sequential patterns of learner actions and utterances associated with outcomes of simulation-based guided discovery learning. We conducted a sequence analysis of data gathered from 11 undergraduate students engaged in discovery learning. Three related methods were used for the sequence analysis: Levenshtein edit distance, k-means clustering of the Levenshtein distance, and the Kohonen generalized median sequence. The median sequences of high-gaining and low-gaining participants showed qualitative differences in how they gathered evidence, stated claims, and drew explanatory inferences. Differences between the sequences of actions and utterances of high-gaining and low-gaining participants suggested ways that students might be guided to enhance discovery learning. By tracking the learning patterns of learners, researchers can determine the conditions under which prompts should be provided and offer recommendations for transforming less effective learning strategies to more effective ones.

理解科学现象需要学习者构建因果系统的心智模型。基于模拟的发现学习为学习者提供了构建心理模型并根据模拟行为测试它们的机会。本研究的目的是调查与基于模拟的引导发现学习的结果相关的学习者行为和话语的顺序模式。我们对从 11 名从事发现学习的本科生收集的数据进行了序列分析。三种相关方法用于序列分析：Levenshtein 编辑距离，k-表示 Levenshtein 距离和 Kohonen 广义中值序列的聚类。高收益和低收益参与者的中位数序列在他们如何收集证据、陈述主张和得出解释性推论方面显示出质的差异。高获益参与者和低获益参与者的动作序列和话语之间的差异表明可以指导学生加强发现学习的方法。通过跟踪学习者的学习模式，研究人员可以确定应该提供提示的条件，并提出建议，将不太有效的学习策略转变为更有效的学习策略。