Journal of Science Education and Technology

Scientific creativity plays an essential role in science education as an advanced cognitive ability that inspires students to solve scientific problems inventively. The cultivation of scientific creativity relies heavily on effective assessment. Typically, human raters manually score scientific creativity using the Consensual Assessment Technique (CAT), which is a labor-intensive, time-consuming, and error-prone process. The assessment procedure is susceptible to subjective biases stemming from cognitive prejudice, distractions, fatigue, and fondness, potentially compromising its reliability, consistency, and efficiency. Previous research has sought to mitigate these risks by automating the assessment through latent semantic analysis and artificial intelligence. In this study, we developed machine learning (ML) models based on a training dataset that included output labels from the Scientific Creativity Test (SCT) evaluated by human experts, along with input features derived from objectively measurable semantic network parameters (representing the scientific knowledge structure) and eye-tracking blink duration (indicating attention patterns during the SCT). Most models achieve over 90% accuracy in predicting the scientific creativity levels of new individuals outside the training set, with some models achieving perfect accuracy. The results indicate that the ML models effectively capture the underlying relationship between scientific knowledge, eye movements, and scientific creativity. These models enable the fairly objective prediction of scientific creativity levels based on semantic network parameters and blink durations during the SCT, eliminating the need for ongoing human scoring. Therefore, laborious and complex manual assessment methods typically used for SCT can be avoided. This new method improves the efficiency of scientific creativity assessment by automating processes, minimizing subjectivity, providing rapid feedback, and enabling large-scale evaluations, all while reducing evaluators’ workloads.

Intelligent tutoring system (ITS) provides learners with step-by-step problem-solving support through scaffolding. Most ITSs have been developed in the USA and incorporate American instructional strategies. How do non-American students perceive and use ITS with different native problem-solving strategies? The present study compares Stoich Tutor, an ITS with a high level of scaffolding, with ORCCA, an ITS with dynamic scaffolds that can support a range of problem-solving strategies. We conducted a think-aloud study with university students in the USA (N = 10) and Germany (N = 11), where students worked with either Stoich Tutor and ORCCA before solving stoichiometry problems on paper. Two human coders derived a coding scheme to investigate the strategies American and German students employ during problem solving on paper without instructional support. We derive a taxonomy of three stoichiometry problem-solving strategies. Next to the American factor labeling method, this taxonomy includes a strategy based on equation transformations and a previously undocumented strategy using abstract symbols to isolate a target variable and then pluck in given values and compute the solution. German students exclusively used the latter strategy, which was not explicitly supported by any of the two tutoring systems. Further, students who did not use the factor-label method for paper-based problem solving, most of whom were German, initially had difficulty setting appropriate goals and working with fractions in the Stoich Tutor. While German students preferred ORCCA based on short interviews, they more often successfully solved problems in Stoich Tutor. Therefore, Stoich Tutor, although misaligned with German instruction, could still support German students’ learning. Still, revisions to ITS based on local instructional cultures could make them potentially more effective and aligned with curricular goals.

Mobile learning (M-Learning) allows chemistry students to access educational resources anytime, anywhere, aiding in problem-solving and critical thinking skills. However, it is necessary to identify the components and objectives of M-Learning that have specific usage in chemistry education. Despite the global focus on e-learning, there is currently no framework for components and objectives of M-Learning usage in chemistry education that could be used as a basis for implementing M-Learning in chemistry education by teachers and educators. The primary objective of this study is to address this research gap. This study attempted to develop a framework in which the relationship between latent and observed variables for the components and objectives of M-Learning usage in chemistry education is examined. The statistical population of this study consisted of 950 teacher-students. The sample size was determined using the Morgan table and a random probability sampling method. Data were collected using a researcher-made questionnaire. The questionnaire was validated by an expert group and found to have a reliability coefficient of 0.837 measured by Cronbach’s alpha. By employing exploratory factor analysis, we validated and categorized 48 objectives into 6 components: knowledge, skill, attitude, technology, tool, and responsibility. Based on the confirmatory factor analysis, we confirmed the fit indexes. The developed framework in this research can be used to develop targeted interventions that address the most critical areas of M-Learning usage and provide a unique and engaging learning experience for chemistry students. Our results have important implications for the design and implementation of mobile learner-centered learning strategies.

Computational thinking (CT) has gained more value for individuals in a world reshaped by digital transformation in the last decade. Therefore, educators and researchers are trying to integrate CT into teaching practices. Efforts to teach CT are increasing, especially in basic courses widely included in school curricula. The focus of this study is the integration of CT into science teaching in the flipped classroom model. In this context, the effects of flipped computational science laboratory (Flipped-CSL) activities carried out with teacher candidates on CT skills, laboratory entrepreneurship, and attitude were investigated. An intertwined mixed research design, in which quantitative and qualitative data were evaluated together, was used in the study. Findings showed that flipped-CSL activities were effective for teacher candidates and improved their CT skills, laboratory entrepreneurship, and attitudes significantly and positively. The results of this study include the practical use of flipped-CSL activities when planning laboratory activities for school science subjects to improve CT skills. Implications for using of flipped-CSL activities in science education were discussed, and suggestions were made regarding the results.

This study explored the impact of Argumentation-driven inquiry (ADI) on primary school students’ argumentation performance in a blended synchronous learning environment (BSLE). A total of 159 fifth-grade primary school students (79 from an urban school and 80 from a rural school) participated in this quasi-experimental study. Students in the control group received inquiry-based (without argumentation) instruction in BSLE, while the experimental group students were instructed in an ADI approach in BSLE. Argumentation performance was measured before and immediately after the four-week research intervention. Results demonstrated that the remote (rural) school students in the experimental group scored significantly higher than remote (rural) school students in the control group on the Claim and Evidence dimensions, while no significant difference existed on the Reasoning and Counterclaim dimensions. The experimental group’s onsite (urban) school students scored significantly higher on the Claim and Reasoning dimensions than the control group’s onsite (urban) school students, though no significance was found in the overall score. This study also found that within the experimental group, the remote (rural) students scored significantly higher on the Evidence dimension of argumentation than the onsite (urban) students. The results of this study showed that in a BSLE, the ADI instructional approach had positive influence on the argumentation performance of primary school students in both rural and urban sites. Implementing ADI in BSLE could bridge the rural–urban gap in education and promote educational equity.

As science education scholars learn more about how people learn, instructors have begun to shift from teaching science as lists of facts and asking students to synthesize ideas into cognitive models or networks. Therefore, the methodologies we use to understand students’ and instructors’ ways of knowing need to capture this complexity. Within education, one methodology that has emerged to capture this complexity is epistemic network analysis (ENA). ENA is a potentially useful tool for understanding connections between people’s ideas and cognitive constructs. Because of its mixed methods approach, ENA is able to provide the depth of qualitative analysis and allow synthesis and comparison across large quantities of data. In this review, we present findings from a scoping literature review of ENA in science education. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 framework, we extracted data from 19 articles. This data consisted of both context-related variables (i.e., disciplinary field) and application-based variables (i.e., theoretical frameworks, research design). Finally, we discuss the findings from this review and their implications for science education.

This article highlights the challenges faced in educating pre-service science teachers in a post-truth era informed by the widespread presence of Generative Artificial Intelligence (GenAI). It examines a significant challenge associated with integrating artificial intelligence into educational methods, where the consequences of GenAI’s tendency to generate hallucinations highlight tensions around the fundamental nature of truth in the field of Science Education. The paper argues that the conventional linear model of knowledge creation and dissemination is inadequate for addressing the complexities linked to evaluating and instructing science in an era marked by the prevalence of generated ‘facts’. By advocating for the educating of science teachers through the conceptual lens of technical agonism, the article presents a framework for the emerging skills that pre-service science teachers need to navigate advancements in GenAI. Importantly, it initiates a discussion on the evolving meanings of ‘objectivity’ and ‘evidence’ in Science Education within a post-truth age shaped by GenAI drawing on Manne’s Logic of Misogyny, laying the groundwork for addressing emerging inquiries into the understanding of ‘truth’ in Science Education influenced by GenAI.

The aim of the research was to explore the effect of educational interventions implemented in desktop-based VR and ımmersive virtual reality (IVR) settings on gifted students’ nature relatedness and attitudes towards environment. A a single group pretest-posttest weak experimental model was implemented. The participants involved in the study consisted of a group of 27 gifted students (age mean = 12.8), comprising 10 males and 17 females. The training program was implemented across three distinct sessions utilizing the Imedu VR platform. The training program used Cognitive Theory of Multimedia Learning principles, applying Exaggerated Feedback in the second session and Corrective Feedback in the third as instructional design elements. The initial phase encompassed a duration of 120 minutes of distance learning that emphasized Sustainable Development Goals, with a particular focus on ocean acidification. Subsequently, participants engaged in a 30-minute IVR experience utilizing Oculus Quest 2 headsets to investigate the effect of ocean acidification on organisms and ecosystems. The final session comprised a 30-minute escape room activity conducted in a distance learning format via Imedu to reinforce what the students learned. Preceding and succeeding the intervention, students completed the Nature Relatedness (NR) Scale and the Revised New Environmental Paradigm Scale (R-NEP). Following confirmation of data normality, a paired samples t-test was employed to compare pretest and posttest scores. The study revealed a statistically significant enhancement in nature relatedness and attitudes towards environment among gifted students (p < .05). Effect sizes were high for NR (Cohen’s d = 1.099) and moderate for R-NEP (Cohen’s d = 0.539). The integration of VR technologies in gifted education for cultivating nature relatedness and attitudes towards environment is thus encouraged.

This design-based research investigated the differences in knowledge-creation practices among university students across two blended learning formats—traditional and hybrid—over multiple years, guided by knowledge-building principles. Each year, 74 first-year students participated in the course to develop new happiness indices through small-group activities. They used a Computer-Supported Collaborative Learning (CSCL) system to share weekly reflection notes and plan future activities in addition to face-to-face interactions. In the hybrid blended learning format, students had to manage communication with remote group members. The students’ reflection notes in CSCL were analyzed to evaluate the development of their epistemic views and perceptions of the community of inquiry (CoI). Clustering analysis revealed that in the traditional blended learning year, 57% of successful students developed their epistemic views over the modules, while 43% were only partially successful. In the hybrid blended learning year, only 36% of students were partially successful, and 64% were not successful. Furthermore, epistemic network analysis (ENA) indicated that students in the traditional blended learning year emphasized cognitive presence, whereas those in the hybrid blended learning year focused more on social presence. These findings suggest that hybrid blended learning should incorporate multimodal communication to reduce cognitive load and enhance epistemic engagement.

This scoping review explores the intersection of climate change education and technology usage in the classroom. We conducted a scoping review of 93 articles to ascertain what, if any, trends exist across this body of scholarship. Findings indicate that technology serves to provide learners with refutation experiences for accepting and making meaning of climate change. While a wide range of technologies are being engaged in formal learning to support learning around climate change, specific technologies are of mixed value to teachers and students. Furthermore, data science skills and competencies play a central role in most technological climate change learning experiences.

Evolution is a key biological concept, and natural selection is an important mechanism of evolution, but studies indicate students reason about natural selection differently based on organismal context. This paper investigates students’ explanations of natural selection in varying contexts after a computational thinking (CT)-central unit designed to scaffold natural selection transfer. The research questions address natural selection change, contextual differences in students’ explanations, and patterns of cooccurrences in students’ natural selection explanations. Students learned about natural selection through scaffolded transfer via Computational Thinking through Algorithmic Explanations (CT-AE), an unplugged instructional approach. The data source is students’ explanations of four pre- and post-unit natural selection scenarios about bacteria, mice, lilies, and mosquitos. This mixed methods study included nonparametric statistics to determine differences between contexts in post-unit explanations and Epistemic Network Analysis (ENA) to create and compare networks of co-occurrences in students’ explanations. There were significant differences between the four pre-unit scenario explanations, but the post-unit explanations displayed fewer differences. ENA analysis indicated that student responses for each scenario were not significantly different. These trends indicate students’ explanations of natural selection based on context varied less after the unit. These results suggest that the unit was successful in scaffolding transfer of natural selection context across contexts.

This study explores the potential to enhance the performance of convolutional neural networks (CNNs) for automated scoring of kinematic graph answers through data augmentation using Deep Convolutional Generative Adversarial Networks (DCGANs). By developing and fine-tuning a DCGAN model to generate high-quality graph images, we explored its effectiveness compared to other models. The augmented dataset was used to train the CNN model, and k-fold cross-validation demonstrated performance improvements. To ensure data quality, defective images generated by the DCGAN were identified and filtered using the CNN model before augmentation. We compared two approaches—augmenting data with and without filtering—and found superior performance when defective images were removed. Additionally, an analysis of the amount of augmented data used revealed a point at which further augmentation no longer significantly improved performance. These findings underscore the importance of specialized DCGAN models and careful dataset curation in improving automated graph scoring systems.

Engineering design that requires mathematical analysis, scientific understanding, and technology is critical for preparing students for solving engineering problems. In simulated design environments, students are expected to learn about science and engineering through their design. However, there is a lack of understanding concerning linking science concepts with design problems to design artifacts. This study investigated how 99 high school students applied science concepts to solarize their school using a computer-aided engineering design software, aiming to explore the interaction between students’ science concepts and engineering design behaviors. Students were assigned to three groups based on their design performance: the achieving group, proficient group, and emerging group. By mining log activities, we explored the interactions among students’ application of science concepts, engineering design behaviors, design iterations, and their design performance. We found that the achieving group has a statistically higher number of design iterations than the other two performance groups. We also identified distinctive transition patterns in students’ applying science concepts and exercising design behaviors among three groups. The implications of this study are then discussed.

This study examines the effect of technology supported guided inquiry environmental education (TsGIEE) on preservice science teachers’ attitudes towards sustainable environment (AtSE), environmental education self-efficacy (EESE) and digital literacy (DL). Embedded design from mixed research method was preferred in the study. 33 (20 female-13 male) fourth grade preservice science teachers participated in the study. A significant difference was found between the pre-test and post-test scores obtained from the overall scale of EESE. In addition, a significant difference was found between the scores obtained from the content knowledge (CK) and instructional strategies (IS) sub-dimensions of the EESE in favor of the post-test. Moreover, a significant difference in favor of the post-test was found between the scores obtained from the overall AtES scale and the use of environmental resources (UER) sub-dimension. There was no significant difference between the scores obtained from the general and sub-dimensions of the digital literacy (DL) scale. When the qualitative data were analyzed, the participants produced metaphors in the category of raising environmental awareness the most regarding TsGIEE. In addition, the participants emphasized that the applications contained informative and useful information and that they gained competence in teaching environmental issues thanks to these applications. Within the framework of the findings obtained, it is recommended to organize technology-supported guided inquiry learning environments in environmental education.

Elementary teachers require support through professional learning activities to enhance their climate change literacy and bolster their self-efficacy for teaching climate change. This study explores methods for supporting in-service elementary teachers’ self-efficacy in climate change teaching by examining the impact of professional learning activities that incorporate learning technologies on climate change literacy. We present the findings from two in-service elementary teachers’ perspectives on how learning technologies facilitated the scaffolding of their self-efficacy for teaching climate change. Initially, both teachers held non-normative beliefs that ozone depletion was a cause of climate change. Interactive visualizations of climate models and simulations of the greenhouse effect enriched the teachers’ climate change literacy, leading to a correct understanding of the causal mechanisms linking greenhouse gases, energy, and global temperatures. The experience also elevated the teachers’ self-efficacy for teaching science concepts, with a particular focus on their understanding of the greenhouse effect and climate change.

As students read scientific texts created in generative artificial intelligence (GenAI) tools, they need to draw on their epistemic knowledge of GenAI as well as that of science. However, only a few research discussed multimodality as a methodological approach in characterising students’ ideas of GenAI-science epistemic reading. This study qualitatively explored 44 eighth and ninth graders’ multimodal representations of ideas about GenAI-science epistemic reading and developed an analytical framework based on Lemke’s (1998) typology of representational meaning, namely presentational, organisational, and orientational meanings. Under each representational meaning, several categories were inductively generated while students expressed preferences in using drawn, written, or both drawn and written mode to express certain categories. Findings indicate that a multimodal approach is fruitful in characterising students’ semiotic resources in meaning-making of ideas about GenAI-science epistemic reading. We suggested implications regarding future intervention studies on tracking students’ ideas about GenAI-science epistemic reading using the analytical framework developed in this study.

Recent advances in generative artificial intelligence are expected to change jobs and education dramatically. This study explores Korean physics educators’ perceptions of the educational use of ChatGPT and future changes in society and the educational environment. For this purpose, semi-structured interviews were conducted with ten Korean physics teachers from high schools and universities, and the interview data were analyzed using consensus qualitative research methods, yielding six categories and 34 subcategories. The results are as follows: First, educators reported the strengths and weaknesses of ChatGPT in physics problem-solving. Second, while ChatGPT has the potential to enhance personalized, learner-led, inquiry-based education, comprehensive efforts across the curriculum, teaching, assessment, infrastructure, and learning communities will be required. Lastly, the rapid progress of artificial intelligence will entail significant changes like shifts in educational paradigms, widening educational inequalities, and reduced teacher influence, necessitating the enhancement of teachers’ digital competencies. This study offers insights to guide future physics and science education by presenting current and upcoming challenges.

According to self-determination theory, frustration of basic psychological needs (autonomy, competence, and relatedness) leads to ill-being and negatively affects the learning process. The present study aimed to analyze the effects of a gamified creativity-based teaching method of human anatomy on basic psychological needs frustration compared with a conventional teaching method. A quasi-experimental design was employed, comparing two anatomy educational treatments (experimental and control) over a 7-week period. A total of 116 first-year students from two Spanish public universities was participated. The Basic Psychological Needs Frustration Scale was utilized, and pre- and post-treatment measurements were collected. Statistical analyses included independent samples t-tests, one-way ANCOVAs, and a factorial repeated measures ANCOVA 2 × 2 (time × treatment), comparing two groups based on time (baseline vs. follow-up) and treatment (control vs. experimental). The analysis revealed that the gamified creativity-based program achieved lower frustration of basic psychological needs compared to the control treatment (t (108) = 3.74, p < .001, d = 0.68) and a treatment effect was observed (F (1) = 9.06, p = .003, η²p = .083). Autonomy and competence frustration significantly increased in the control group, while apparently remained consistent over time in the treatment group. Baseline and follow-up significant differences were found for relatedness (t (114) = 1.12, p = .03, d = 0.4; t (110) = 2.88, p = .005, d = 0.53, respectively), as well as a treatment effect (F (1) = 7.28, p = .008, η²p = .069). These findings provide support for the idea that students’ basic psychological needs are lower frustrated with the implementation of a gamified creativity-based teaching method in anatomy education.

Results of a Principal Investigators Programmatic Data Inventory (PDI) of a National Science Foundation Robert Noyce Track Four project are discussed in this paper. The PDI results shed light on the development of STEM teacher scholars as they progress through the programs and of the qualifications and procedures of the application process. The PDI inventory focused on the demographics and qualifications of the Noyce scholar pool, changes during the training process, progress of high-priority teacher candidate groups, etc. A survey was developed and administered anonymously through Qualtrics to the Principal Investigators in four institutions housing Noyce scholarship programs. Key findings of the Inventory are presented and discussed with recommendations and implications.

The emergence of AI chatbots brings new opportunities for active and personalized learning. Integrating AI chatbots into instruction requires guidance from the teacher to scaffold students’ purposeful scientific discovery. The present study framed the AI chatbot within guided discovery learning (GDL) and integrated the ARCS motivational model to create an innovative instructional strategy. The impact of this AI chatbot-supported GDL approach on pre-service teachers’ learning performance and motivation in global warming education was investigated. A pre-experimental design was employed with 59 sophomore teacher education majors at a public university in southern China. Data were collected through pre- and post-tests of global warming knowledge, an ARCS model survey, and focus group interviews. Results showed a significant improvement in students’ global warming knowledge and positive motivation across all ARCS categories. Qualitative analysis revealed that the approach enhanced students’ attention, perceived relevance, confidence, and satisfaction in learning about global warming. The integration of AI chatbots with GDL facilitated active engagement, personalized learning, and the development of critical thinking skills. The findings suggest that combining emerging AI technologies with established pedagogical frameworks can create engaging and effective learning experiences in environmental education. This study contributes to the potential of AI-supported learning approaches to empower future educators in addressing global environmental challenges.

This paper explores the transformative impact of generative artificial intelligence (GenAI) on teachers’ roles and agencies in education, presenting a comprehensive framework that addresses teachers’ perceptions, knowledge, acceptance, and practices of GenAI. As GenAI technologies, such as ChatGPT, become increasingly integrated into educational settings, both in-service and future teachers are required to adapt to evolving classroom dynamics, where AI plays a significant role in content creation, personalized learning, and student engagement. However, existing literature often treats these factors in isolation, overlooking how they collectively influence teachers’ ability to effectively integrate GenAI into their pedagogical practices. This paper fills this gap by proposing a framework that categorizes teachers (including both pre- and in-service teachers, hereafter) into four roles—Observer, Adopter, Collaborator, and Innovator—each representing different levels of GenAI engagement, outlining teachers’ agencies in GenAI classrooms. By highlighting the need for quality teacher education programs, continuous professional development and institutional support, we use examples to demonstrate how teachers can evolve from basic GenAI users to co-creators of knowledge alongside GenAI systems. The findings emphasize that for GenAI to reach its full educational potential, teachers must not only accept and understand its capabilities but also integrate it deeply into their teaching practices. This study contributes to the growing literature on GenAI in education, offering practical implications for supporting both in-service and future teachers in navigating the complexities of GenAI adoption.

Integrating generative artificial intelligence (GenAI) in pre-service teachers’ education programs offers a transformative opportunity to enhance the pedagogical development of future science educators. This conceptual paper suggests applying the GenAI tool to evaluate pedagogical content knowledge (PCK) among pre-service science teachers. By holding interactive dialogues with GenAI, pre-service teachers engage in lesson planning in a way that reveals their understanding of content, pedagogy, and PCK while facilitating the practical application of theoretical knowledge. Interpretation of these interactions provides insights into teachers-to-be knowledge and skills, enabling personalized learning experiences and targeted program adjustments. The paper underscores the need to equip pre-service teachers with the necessary competencies to utilize GenAI effectively in their future teaching practices. It contributes to the ongoing discourse on technology’s role in teacher preparation programs, highlighting the potential of addressing existing challenges in evaluating and developing teacher knowledge via GenAI. The suggested future research directions aim to further investigate the GenAI usage implications in educational contexts.

The sudden transition to remote learning in response to the COVID-19 pandemic resulted in a significant increase in reliance on instructional technology, some unique to STEM disciplines. This violation of students’ expectations for the learning experience presented an additional crisis. Students had to react to the disruption without time to carefully consider how best to use new technologies for learning. In crises, sensemaking occurs retroactively; students made sense of new instructional technologies after they had navigated the crisis. They adopted new approaches to engaging with their courses based on what strategies seemed successful during the pandemic. In this article, we use sensemaking as a lens to explore the changes in students’ views of instructional technologies as a result of their experiences during the pandemic. We conducted focus groups with 31 undergraduate STEM students at a medium-sized public university in the southeastern US. We found that STEM students were initially overwhelmed with the amount of technology they were required to use in their courses, but that in the long term, they expect more technology after the pandemic, especially to facilitate flexibility in learning space and approach. While they found some new tools that seemed to support their learning, they also developed ineffective strategies (e.g., navigating e-textbooks to find key words rather than reading). The results of our study inform post-COVID best practices for incorporating instructional technology. Faculty should expect to have to teach students how to use technology rather than assuming they will learn naturally. Faculty will also need to consider that while technology affords some support for learning, it introduces opportunities for students to employ strategies that interfere with learning.

High-quality science assessments are multi-dimensional. They promote disciplinary practices, core ideas, cross-cutting concepts, and science sense-making. In this paper, we investigate the feasibility of using generative artificial intelligence (GenAI), specifically multimodal large language models (MLLMs), to annotate and provide improvement ideas for K-12 science assessments. The AI-generated annotations critique how the assessments align with the three dimensions of the Next Generation Science Standards (NGSS) and suggest ideas to elicit students’ science sense-making. We outline our process with various prompting strategies: few-shot and zero-shot learning (prompting with and without examples), chain of thought (eliciting the MLLM’s reasoning), and sampling strategies (outputting high or low level of randomness). Overall, the AI annotations can reason about the alignment between the assessments and NGSS dimensions and overlap with annotations from K-12 educators. Annotations generated with few-shot learning generally score higher overall and provide more details than zero-shot prompts. Further, interviews with science teachers reveal that the MLLM annotations can support teachers’ reflection on instructional practices and assessment revision. We discuss the application of MLLMs to develop three-dimensional science assessments.

The current data science education requires educators to provide more personally meaningful and culturally relevant data science learning experiences. By incorporating art production, we designed a data-art inquiry program to teach students data science basics and enable them to use art techniques to visualize their data. To understand how students established connections between data and art, we employed epistemic network analysis (ENA) at two levels to explore how they combine their data practices and art processes. Our study had three primary findings: (1) students have a personalized way of combining data and art; (2) data collection appears to be a key practice in our data-art inquiry program; (3) art production empowers students to better understand and communicate their data. In the future design of a data-art inquiry program, we suggest programming might (1) encourage students to explore personalized art formats to present their data, (2) emphasize the role of data collection, and (3) provide sufficient space and time for art processes to ensure the connections between data and artwork.