Search for: All records

Abstract BackgroundSelf-regulated learning (SRL) strategies can be domain specific. However, it remains unclear whether this specificity extends to different subtopics within a single subject domain. In this study, we collected data from 210 college students engaged in a computer-based learning environment to examine the heterogeneous manifestations of learning behaviors across four distinct subtopics in introductory statistics. Further, we explore how the time spent engaging in metacognitive strategies correlated with learning gain in those subtopics. ResultsBy employing two different analytical approaches that combine data-driven learning analytics (i.e., sequential pattern mining in this case), and theory-informed methods (i.e., coherence analysis), we discovered significant variability in the frequency of learning patterns that are potentially associated with SRL-relevant strategies across four subtopics. In a subtopic related to calculations, engagement in coherent quizzes (i.e., a type of metacognitive strategy) was found to be significantly less related to learning gains compared to other subtopics. Additionally, we found that students with different levels of prior knowledge and learning gains demonstrated varying degrees of engagement in learning patterns in an SRL context. ConclusionThe findings imply that the use—and the effectiveness—of learning patterns that are potentially associated with SRL-relevant strategies varies not only across contexts and domains, but even across different subtopics within a single subject. This underscores the importance of personalized, context-aware SRL training interventions in computer-based learning environments, which could significantly enhance learning outcomes by addressing the heterogeneous relationships between SRL activities and outcomes. Further, we suggest theoretical implications of subtopic-specific heterogeneity within the context of various SRL models. Understanding SRL heterogeneity enhances these theories, offering more nuanced insights into learners’ metacognitive strategies across different subtopics. 

Learning online is now ubiquitous. However, teachers’ self-directed and guided learning online deserves further exploration because most research on successful teacher professional learning has been conducted on in-person programs. The present study examined teacher behaviors in an online platform designed to support teachers’ professional learning in elementary mathematics. In particular, this study explored whether teacher behaviors on an online professional learning platform clustered in ways that suggest distinct use cases and whether those behaviors were associated with particular teacher characteristics. Results revealed a cluster of teachers who predominately focus their behaviors on the guided learning modules on the website, which was associated with teacher characteristics, including being less likely to enjoy teaching mathematics and being newer to teaching the curriculum supported on the website. Implications for future research and for supporting teacher learning are discussed. 

Educational data mining has allowed for large improvements in educational outcomes and understanding of educational processes. However, there remains a constant tension between educational data mining advances and protecting student privacy while using educational datasets. Publicly available datasets have facilitated numerous research projects while striving to preserve student privacy via strict anonymization protocols (e.g., k-anonymity); however, little is known about the relationship between anonymization and utility of educational datasets for downstream educational data mining tasks, nor how anonymization processes might be improved for such tasks. We provide a framework for strictly anonymizing educational datasets with a focus on improving downstream performance in common tasks such as student outcome prediction. We evaluate our anonymization framework on five diverse educational datasets with machine learning-based downstream task examples to demonstrate both the effect of anonymization and our means to improve it. Our method improves downstream machine learning accuracy versus baseline data anonymization by 30.59%, on average, by guiding the anonymization process toward strategies that anonymize the least important information while leaving the most valuable information intact. 

How universities can influence socially responsible AI technology development and use. 

While classroom video data are detailed sources for mining student learning insights, their complex and unstructured nature makes them less than straightforward for researchers to analyze. In this paper, we compared the differences between the processes of expert- informed manual feature engineering and automated feature engi- neering using positional data for predicting student group interac- tion in four middle school and high school mathematics classroom videos. Our results highlighted notable differences, including im- proved model accuracy for the combined (manual features + au- tomated features) models compared to the only-manual-features models (mean AUC = .778 vs. .706) at the cost of feature interpretabil- ity, increased number of features for automated feature engineering (1523 vs. 178), and engineering approach (domain-agnostic in au- tomated vs. domain-knowledge-informed in manual). We carried out feature importance analyses and discuss the implications of the results for potentially augmenting human perspectives about quali- tatively coding classroom video data by confirming and expanding views on which body areas and characteristics may be relevant to the target interaction behavior. Lastly, we discuss our study’s limitations and future work.