This topic spotlight highlights ECR:Core and ECR:BCSER projects that use artificial intelligence (AI) and other computational methods for data analysis in STEM education research.

As educational data become increasingly complex, AI approaches such as natural language processing (NLP) and machine learning (ML) provide rigorous methods for addressing complex educational data. Network-based approaches are being used to analyze STEM learning.

Five themes characterize how ECR projects use AI and other computational methods, as described below. The publications from ECR-funded projects relate to each theme. These examples illustrate the breadth of research within each area.

Across the ECR portfolio, ECR researchers should consider connections between their work and opportunities to apply AI and other computational methodologies.

Topic Themes

AI for Evaluation and Feedback. These methodologies use AI to automate the evaluation of complex student work or instructional pedagogy and provide scalable, real-time feedback loops.

Using Automated Analysis to Assess Middle School Students' Competence with Scientific Argumentation

AI-Assisted Activity Detection in K-6 Classroom Environments: A Preliminary Framework to Assist in Pedagogical Performance Evaluation

"Learning Recorder" that Helps Lesson Study of Collaborative Learning

A Student-Teacher Multimodal Interaction Analysis System For Classroom Observation

Analyzing Response Times And Answer Feedback Tags In An Adaptive Assessment

Analyzing Users' Interaction with Writing Feedback and Their Effects on Writing Performance

Predictive & Cognitive Modeling. Research in this category employs computational models to simulate cognitive processes—such as self-regulation—or forecast long-term student outcomes based on interaction data.

Representing and Predicting Student Navigational Pathways in Online College Courses

Comparison of Machine Learning Performance Using Analytic and Holistic Coding Approaches Across Constructed Response Assessments Aligned to a Science Learning Progression

Transfer from Spatial Education to Verbal Reasoning and Prediction of Transfer from Learning-Related Neural Change

Data Augmentation With GAN To Improve The Prediction Of At-Risk Students In A Virtual Learning Environment

Network & Relational Analysis. These projects highlight methods to map the complex structure of relationships, especially to study how interpersonal (peer/mentor) interactions and intrapersonal knowledge networks influence participation and persistence in STEM.

Mapping the Creative Personality: A Psychometric Network Analysis of Highly Creative Artists and Scientists

Cracking the Code of Learning Gains: Using Ordered Network Analysis to Understand the Influence of Prior Knowledge

Cognitive, Affective, and Politicized Trust in a Community Youth Program: A Participatory Design Research Project

Multimodal Learning Analytics. These projects showcase approaches to process and/or triangulate data from non-textual sources, such as video, audio, and biosensors, to construct a holistic view of learner engagement in real-world environments. Studies in this subtopic typically analyze the physical and physiological dimensions of learning that traditional methods often miss.

Spoken Interaction Modeling for Automatic Assessment of Collaborative Learning

Cortical Cognitive Processing During Reading Captured Using Functional-Near Infrared Spectroscopy

What Eye Movements Reveal About Later Comprehension of Long Connected Texts

Tracking Individuals in Classroom Videos via Post-Processing Openpose Data

Prediction Of Students' Self-Confidence Using Multimodal Features In An Experiential Nurse Training Environment

Evaluating AI as a Methodology. These projects focus on the validation, interpretability, and ethical auditing of methods themselves to ensure that algorithmic findings are psychometrically sound before being applied to broader populations. 

A Meta-Analysis of Machine Learning-Based Science Assessments: Factors Impacting Machine-Human Score Agreements

Can AI be Racist? Color-Evasiveness in the Application of Machine Learning to Science Assessments

From Substitution to Redefinition: A Framework of Machine Learning-Based Science Assessment

On the Validity of Machine Learning-Based Next Generation Science Assessments: A Validity Inferential Network

Trust in Generative AI Among Students

Note on our methodological approach. To generate the existing list of topics, we applied topic modeling to available publications from ECR‑funded projects. Topic modeling is a text mining technique that uses NLP on large sets of texts to explore themes. The results of our topic modeling surfaced multiple overlapping and non-overlapping themes which were then reviewed and organized into thematic groups. The publications shown on this page are examples drawn from these thematic groups.

Photo by Allison Shelley/Complete College Photo Library