[This paper is part of the Focused Collection on Qualitative Methods in PER: A Critical Examination.] Qualitative research methods have provided key insights in physics education research (PER) by drawing on non-numerical data such as text or video data. While different methods towards qualitative research exist, they share two essential steps: recognizing patterns in the data and interpreting these patterns. Although these methods have led to the development of rigorous theory, there are challenges: As such methods require a series of judgments by the analyst, they are difficult to validate and reproduce. Further, they are hard to scale so that they are unavailable to the analysis of large-scale data. In this way, important phenomena may remain inaccessible to qualitative analysis. Reacting to these challenges and leveraging the potential of emerging methods of artificial intelligence (AI) such as machine learning and natural language processing, sociologist Nelson has proposed the concept of computational grounded theory (CGT). CGT proceeds in a process of three consecutive steps: In the first step, one leverages the power of computational techniques, especially natural language processing and unsupervised machine learning techniques, for pattern detection in large datasets—those of a size and scope that may prohibit human-driven analysis from the outset. In the second step, one relies on the integrative and interpretative capabilities of human researchers to add quality and depth to the quantity and breadth of the first step. In the last step, one again uses computational techniques to test the extent to which the detected and refined patterns from the first and second step hold throughout the whole dataset under investigation. Interestingly, CGT does not aim at simply automating parts of the qualitative process by using AI, but rather aims at integrating AI into the human analyst’s workflow within a qualitative analysis. This leads to an analytical system that can do something that is quantitatively and qualitatively different from what a human or machine can do alone. In this way, CGT aims at addressing questions about validity, reproducibility, and scalability in qualitative research while preserving the theoretical sensitivity and unique inferencing capabilities of the human analyst. In this paper, we provide a primer on CGT, present how it can be used to investigate the physics problem-solving approaches of $N=417$ students based on textual data, and discuss CGT’s potentials and challenges in PER. In consequence, this paper can provide critical input to the discussion of how emerging AI technologies can provide new avenues in qualitative PER.

National educational standards stress the importance of science and mathematics learning for today’s students. However, across disciplines, students frequently struggle to meet learning goals about core concepts like energy. Digital learning environments enhanced with artificial intelligence hold the promise to address this issue by providing individualized instruction and support for students at scale. Scaffolding and feedback, for example, are both most effective when tailored to students’ needs. Providing individualized instruction requires continuous assessment of students’ individual knowledge, abilities, and skills in a way that is meaningful for providing tailored support and planning further instruction. While continuously assessing individual students’ science and mathematics learning is challenging, intelligent tutoring systems show that it is feasible in principle. However, the learning environments in intelligent tutoring systems are typically not compatible with the vision of how effective K-12 science and mathematics learning looks like. This leads to the challenge of designing digital learning environments that allow for both – meaningful science and mathematics learning and the reliable and valid assessment of individual students’ learning. Today, digital devices such as tablets, laptops, or digital measurement systems increasingly enter science and mathematics classrooms. In consequence, students’ learning increasingly produces rich product and process data. Learning Analytics techniques can help to automatically analyze this data in order to obtain insights about individual students’ learning, drawing on general theories of learning and relative to established domain specific models of learning, i.e., learning progressions. We call this approach Learning Progression Analytics (LPA). In this manuscript, building of evidence-centered design (ECD), we develop a framework to guide the development of learning environments that provide meaningful learning activities and data for the automated analysis of individual students’ learning – the basis for LPA and scaling individualized instruction with artificial intelligence.

Background: Formative assessments are needed to enable monitoring how student knowledge develops throughout a unit. Constructed response items which require learners to formulate their own free‐text responses are well suited for testing their active knowledge. However, assessing such constructed responses in an automated fashion is a complex task and requires the application of natural language processing methodology. In this article, we implement and evaluate multiple machine learning models for coding energy knowledge in free‐text responses of German K‐12 students to items in formative science assessments which were conducted during synchronous online learning sessions.Dataset: The dataset we collected for this purpose consists of German constructed responses from 38 different items dealing with aspects of energy such as manifestation and transformation. The units and items were implemented with the help of project‐based pedagogy and evidence‐centered design, and the responses were coded for seven core ideas concerning the manifestation and transformation of energy. The data was collected from students in seventh, eighth and ninth grade.Methodology: We train various transformer‐ and feature‐based models and compare their ability to recognize the respective ideas in students' writing. Moreover, as domain knowledge and its development can be formally modeled through knowledge networks, we evaluate how well the detection of the ideas within responses translated into accurate co‐occurrence‐based knowledge networks. Finally, in terms of the descriptive accuracy of our models, we inspect what features played a role for which prediction outcome and if the models pick up on undesired shortcuts. In addition to this, we analyze how much the models match human coders in what evidence within responses they consider important for their coding decisions.Results: A model based on a modified GBERT‐large can achieve the overall most promising results, although descriptive accuracy varies much more than predictive accuracy for the different ideas assessed. For reasons of comparability, we also evaluate the same machine learning architecture using the SciEntsBank 3‐Way benchmark with an English RoBERTa‐large model, where it achieves state‐of‐the‐art results in two out of three evaluation categories.

Energy is a central concept in science in every discipline and also an essential player in many of the issues facing people everywhere on the globe. However, studies have shown that by the end of K‐12 schooling, most students do not reach the level of understanding required to be able to use energy to make sense of a wide range of phenomena. Many researchers have questioned whether the conceptual foundations of traditional approaches to energy instruction may be responsible for students' difficulties. In response to these concerns, we developed and tested a novel approach to middle school physical science energy instruction that was informed by the recommendations of the Framework for K‐12 Science Education (National Research Council, 2012a) and the Next Generation Science Standards (NGSS) (NGSS Lead States, 2013). This new approach differs substantially from more traditional approaches to energy instruction in that it does not require energy forms and it emphasizes connections between energy, systems, and fields that mediate interaction‐at‐a‐distance. We investigated student learning during this novel approach and contrasted it with student learning within a comparable unit based on a more traditional approach to energy instruction. Our findings indicate that students who learned in the new approach outperformed students who learned in the traditional approach in every quantitative and qualitative aspect considered in this study, irrespective of their prior knowledge of energy. They developed more parsimonious knowledge networks in relation to energy that focused primarily around the concept of energy transfer. This study warrants further investigation into the value of this new approach to energy instruction in both middle and high school.

Machine learning (ML) has become commonplace in educational research and science education research, especially to support assessment efforts. Such applications of machine learning have shown their promise in replicating and scaling human‐driven codes of students' work. Despite this promise, we and other scholars argue that machine learning has not yet achieved its transformational potential. We argue that this is because our field is currently lacking frameworks for supporting creative, principled, and critical endeavors to use machine learning in science education research. To offer considerations for science education researchers' use of ML, we present a framework, Distributing Epistemic Functions and Tasks (DEFT), that highlights the functions and tasks that pertain to generating knowledge that can be carried out by either trained researchers or machine learning algorithms. Such considerations are critical decisions that should occur alongside those about, for instance, the type of data or algorithm used. We apply this framework to two cases, one that exemplifies the cutting‐edge use of machine learning in science education research and another that offers a wholly different means of using machine learning and human‐driven inquiry together. We conclude with strategies for researchers to adopt machine learning and call for the field to rethink how we prepare science education researchers in an era of great advances in computational power and access to machine learning methods.

Uncertainty is ubiquitous in science, but scientific knowledge is often represented to the public and in educational contexts as certain and immutable. This contrast can foster distrust when scientific knowledge develops in a way that people perceive as a reversals, as we have observed during the ongoing COVID-19 pandemic. Drawing on research in statistics, child development, and several studies in science education, we argue that a Bayesian approach can support science learners to make sense of uncertainty. We provide a brief primer on Bayes’ theorem and then describe three ways to make Bayesian reasoning practical in K-12 science education contexts. There are a) using principles informed by Bayes’ theorem that relate to the nature of knowing and knowledge, b) interacting with a web-based application (or widget—Confidence Updater) that makes the calculations needed to apply Bayes’ theorem more practical, and c) adopting strategies for supporting even young learners to engage in Bayesian reasoning. We conclude with directions for future research and sum up how viewing science and scientific knowledge from a Bayesian perspective can build trust in science.