Virtual Exchange (VE), a pedagogical practice in which students from different countries collaborate to complete a series of activities, is a technique that can strengthen students’ skills in several areas, including their ability to collaborate effectively in international teams upon graduation. This paper explores the integration of VE activities in statistics and data science education, highlighting the benefits of intercultural communication and active learning in improving students' data analysis and communication skills. Two VE projects between the University of Florida (UF) and the University of Edinburgh (UE) are detailed. VE project 1 involved business graduate students enrolled in a statistics course and undergraduates in different programs from UE creating and refining survey questions and data visualizations. VE project 2 featured graduate students from both universities presenting their coursework projects on statistical modeling and data visualization. Instructors’ perceptions of the effectiveness of these practices are described in terms of subject learning, communication skills, and the overall quality of the collaborative experience. A thematic analysis of students’ reflections on intercultural competencies and the investigative cycle is also provided. The study highlights the potential of VE activities to enhance statistical learning outcomes and intercultural knowledge competencies in data science education. Findings suggest that VE projects in statistics and data science could benefit from incorporating a broad cultural context and structured feedback mechanisms.
The field of statistics and data science education research needs to grow to meet the demands of the increased importance of statistical literacy to an educated society and workforce. This article describes a progression of inquiry starting at the level of Teaching and moving through to the levels of the Scholarship of Teaching and Education Research. Examples of each level of inquiry are provided from the statistics and data science education literature and suggestions for academics, for example, faculty, graduate students, or instructors, who want to move between inquiry levels are provided. Academics interested in statistics and data science education can use this paper to identify their current level of inquiry, select their preferred level of inquiry, and then use the suggestions provided to move their work into their preferred level of inquiry. The goal of this paper is to inspire more academics to engage in inquiry around the teaching and learning of statistics and data science education to expand and improve statistics and data science instruction on a broader scale.
Given the importance of understanding the specific needs for integrating technology into teaching across distinct educational contexts and disciplines, this study explores aspects related to digital inclusion that influence the use of technology in statistics courses at a sample of Mexican colleges. The study stems from a 2020-2024 national research project that characterizes how college instructors teach and assess statistics. A robust survey of 750 instructors was validated using exploratory items assessed on an ordinal scale with confirmatory factor analysis to identify salient characteristics. Results of this survey suggest that critical aspects affecting digital inclusion fall into two categories: first-order barriers (access to technology in the classroom, lack of free software, limited access to technological training opportunities for college instructors), and second-order barriers, like digital literacy among college instructors and students, including skills in using specific technologies and pedagogical practices supported by technology. Results also identify key factors associated with instructors’ software preferences and their use in college-level statistics courses. Our findings indicate that incorporating technology into undergraduate statistics courses requires support from all education stakeholders, including well-prepared instructors, to meet current digital literacy requirements.
To help understand the effect of Large Language Models (LLMs) on data science practice we examine the extent to which self-reported LLM usage is correlated with the mark that a student received on a final paper in a classroom data science setting. We find some mild evidence from this observational study that LLM usage may be associated with better scores, especially for students who do not natively speak English. Additionally, comparing self-reported usage, there was a considerable increase between the class that occurred in January-April 2024 where 41 per cent of students self-reported extensive LLM usage and the class that occurred in September-December 2024 where 69 per cent reported extensive LLM usage. Despite the classroom setting used for evaluation, the task of interest is similar to the work done by professional data scientists. Our finding suggests the need for more extensive work evaluating how LLMs can be integrated into the data science workflow in a way that provides value in both the classroom and the workplace.
Serving millions of diverse learners across the world, Massive Open Online Courses (MOOCs) remain a relevant delivery approach in higher education and beyond. Nevertheless, MOOC environments create some challenges for optimizing learners’ self-regulated learning skills, resulting in high attrition rates in these educational contexts. Pre-existing anxieties can be particularly problematic in the context of statistics-related courses, which are known to create higher levels of anxiety among learners. Informed by the self-regulated learning theory, this study (approved by the Institutional Review Board) investigated whether the anxiety dimension of the Statistics Anxiety Rating Scale (STARS) had evidence of validity and reliability when used to assess learners’ statistics anxiety in a Massive Open Online Course focusing on power and sample size analysis. Results from both exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) provide evidence of validity and reliability. The results suggest that the anxiety dimension of the STARS instrument could be used as a pre-assessment tool to gauge learners’ initial statistics anxiety. Based on these results, theoretical and practical considerations are discussed.
Calls to incorporate causal reasoning into undergraduate statistics education have grown in recent years, yet resources remain scarce. This paper presents a sequence of five activities designed to help students transition from correlational to causal reasoning using real-world police stops data. The paper’s contribution to statistics education lies in providing an implementable framework integrating practical data analysis, counterfactual reasoning through the Rubin Causal Model, and directed acyclic graphs, offering instructors a comprehensive way to introduce causal inference concepts in applied settings. The activities, which ask the question “Do police officers discriminate against Black drivers?”, were implemented in an undergraduate electives course on quantitative analysis of structural injustice. We used publicly available data, case-based discussion prompts, and scaffolded activities. The course enrolled upperclassmen who have completed prerequisites in regression modeling and R programming. Across 300 students between 2022 and 2025, the instructor’s observations and reflections support the idea that activities improve students’ distinguishing correlation from causation and assessing the limits of observational data. The course also enrolled underrepresented groups in statistics, fostering diverse opinions. Practices for introducing controversial topics in statistics education are discussed. The activity sequence discussed offers one adaptable model for integrating causal reasoning into statistics instruction.
This paper describes the implementation of oral exams, a more authentic and potentially more informative form of assessment than written exams, at a large scale (over 600 students), in the context of undergraduate introductory programming classes. Our study focuses on the experiences of the PhD students who both taught the classes as graduate student instructors (GSIs) and administered these exams, where their experiences were captured using surveys and audio recordings of instructor meetings. For those interested in implementing oral exams at large scales in their statistics and data science classes, and who view the amount of time and effort as worth it, we provide several recommendations based on the GSIs and our experiences. These include acquiring and listening to GSI feedback, using technology appropriately, mastering scheduling, training instructors, and preparing students for the exam experience. Our results are most applicable to GSIs at an introductory level on a large scale and with students with minimal statistical knowledge. Supplementary materials for this article are available online.
An ability to situate data in social and cultural context is critical to a responsible data practice. However, curriculum for introducing statistics and data science students to such skills is often underdeveloped in undergraduate programs. This article motivates and describes a course for introducing statistics and data science students to the principles and methods of data ethnography, demonstrating how qualitative data collection and cultural analysis can be brought to bear on the study of data settings and artifacts. I outline how students in the course are taught to culturally analyze data across different social scales, starting with the design of data organizing infrastructures, moving through the scales of individual data labor and collective data practices, and concluding with the scales of data policy, politics, and discourse. At each scale, I provide examples of how students are taught to apply an ethnographic methodology towards analyzing a data setting or artifact, and I provide examples of student work to demonstrate learning outcomes. Assessing student perceptions of their learning, I argue that practicing ethnography develops skill in recognizing and interpreting the cultural meaning behind taken-for-granted data infrastructures and actions, as well as skill in analyzing the social dimensions and impacts of data work.
In this paper we present an instructional unit that engages students in investigating systemic racism in traffic stops using a dataset from Charlotte, NC and that we have implemented in content and methods classes for pre-service teachers as part of a curriculum development research project. The instructional unit has two main goals: to develop statistical concepts and practices associated with critical statistical literacy, and to normalize discussion of race and racism in STEM classrooms. To help instructors and educational researchers consider how they could translate this unit to their own settings, we provide relevant historical context for race and racism in the U.S. along with pedagogical suggestions for teaching about such issues. We discuss the overall design of the unit and provide examples of student work in relation to the activity, as well as possible extension activities. All materials for the unit including instructors’ guide, pedagogical resources and all activity materials and slides are provided in a permanent online repository for others to use and modify.
Data and statistical findings can be communicated effectively using graphs, often better than could be done with numbers or words. In pretest-posttest designs, graphs can be especially useful for visualizing patterns of change. In this study, we sought to examine the quantity of graphs, how variability and error were visualized, and the visual accessibility of graphs used in statistics education literature. Data came from all articles with a two-timepoint design published in three statistics education research journals (Journal of Statistics and Data Science Education, Statistics Education Research Journal, Teaching Statistics) between 2012 and 2024. Results indicate that over half of all published articles implementing pretest-posttest designs had an associated graph. Variability was frequently depicted using distributional methods, but individual-level data and trajectories were rarely shown, limiting the ability to fully represent change over time. Visual accessibility was often poor: small text size and missing or uninformative alt-text were the most common barriers, and many graphs were inaccessible to individuals with color vision deficiencies or reduced contrast sensitivity. To support improved graphing practices, we developed an open-source Shiny web application that allows users to create publication-ready, accessible graphs for pretest–posttest data and provides guidance on effective visualization practices.
This study explores how undergraduate students from diverse academic backgrounds engage with and perceive the integration of data storytelling (DS) and self-regulated learning (SRL) in the context of introductory data science education. Building on established DS frameworks and extending previously developed data stories, this study focuses on two key aspects: (1) students’ perceptions and perspectives regarding the integration of DS and SRL, and (2) based on the existing four stories, potential improvements that can be made to address students’ challenges and needs to inform similar future narrative designs. Participants engaged with these data stories accompanied by comprehension questions to support active learning, followed by group discussion interviews to gain deeper insight into their experiences. A thematic analysis approach was employed to evaluate students’ perceptions and perspectives. Findings reveal that students appreciated the clarity and effectiveness of the current data story design, while also identifying areas for improvement. Overall, the results suggest that integrating SRL with DS can serve as a promising and inclusive pedagogical strategy, fostering SRL engagement and supporting diverse learners in data science education.
The ASCCR (Attitude-Structure-Content-Communication-Relationship) framework was recently developed to teach collaboration skills to statisticians and data scientists. However, its effectiveness in real-world settings has not yet been systematically evaluated. To assess this, we evaluated novice undergraduate and graduate students' performances in initial collaboration meetings with real domain experts and compared them to an expert collaborator. Using video recordings, rubric scores, and domain expert feedback surveys, we found that novices performed surprisingly well compared to the expert. Specifically, novices scored nearly as well as the expert on the Attitude, Structure, and Relationship components of the ASCCR framework. Although novices did not initially perform as well on the Content or Communication aspects, they were able to close the gap. By the end of the collaboration projects, the novices had higher overall domain expert feedback scores than the expert. The primary implication of our study is that novices can become effective collaborators in a very short time. We discuss our findings' practical implications and provide recommendations for integrating the ASCCR framework into statistics and data science collaboration, consulting, and capstone courses.
We describe a pilot implementation of a classroom activity that introduces machine olfaction, a cutting-edge research topic, into an undergraduate deep learning course, STATS 315, at the University of Michigan. Previously, the course focused on text and image data, but we introduce smell as a new modality to expand students’ learning experiences. By incorporating hands-on activities using carefully selected odorants, we engaged students in the data collection process and highlighted the complexities and variability inherent in sensory data. We also exposed students to graph neural networks and the curated GoodScents-Leffingwell dataset. The module adopted a multidisciplinary approach that combines chemistry, machine learning, and sensory evaluation. This pilot provides initial evidence for the feasibility of incorporating machine olfaction into deep learning instruction in the educational context where we implemented the activity, and it suggests directions for future curricular development. We discuss practical challenges encountered and propose directions for improving and extending this educational activity.
The rapid pace of digital transformation has created a pressing need for data literacy. While its importance is widely recognized, there is no clear consensus on the specific skills required. Current Data Literacy Education (DLE) approaches in data science and machine learning courses often focus on digital skills such as data preparation and analysis, overlooking essential foundational skills like creativity, agility, and collaboration. Moreover, as DLE spans beyond traditional STEM boundaries, there is a growing need for inclusive approaches that engage non-STEM learners and address students' diverse needs, interests, and goals. In response, we propose a practical framework for DLE in data science and machine learning courses that cultivates both digital and foundational skills within an inclusive learning environment. Our approach prioritizes active learning and coaching, leveraging our web application STATY to bridge the gap between theory and practice by engaging students in real-world data problem-solving. We illustrate its application through the master's-level course "Business Data Science-From Data to Forecasts and Decisions," along with our experiences and lessons learned. Supplementary materials for this article are available online.
Statistics anxiety is a specific form of anxiety that arises when individuals encounter statistical concepts. It has been shown to affect educational outcomes and hinder academic performance. This study investigated the construct validity, measurement invariance, and discriminant validity of the Hungarian version of the Statistics Anxiety Rating Scale (STARS). Furthermore, we examined the relationship between statistics anxiety and performance in a statistics course and on the Cognitive Reflection Test that measures analytical thinking. The sample comprised 377 participants (85% female) with a mean age of 20.6 years, predominantly freshmen from E & ouml;tv & ouml;s Lor & aacute;nd University and the University of P & eacute;cs. Our findings provide evidence for the reliability and validity of the Hungarian STARS, preserving its bifactor structure and subscales. All levels of measurement invariance were achieved, indicating that the Hungarian STARS reliably assesses statistics anxiety across universities and statistics course experience. Discriminant validity analysis revealed that statistics anxiety is distinct from other constructs. The global level of statistics anxiety had a weak negative correlation with performance and with analytical thinking. These results highlight the reliability and validity of the Hungarian STARS for measuring statistics anxiety in university settings. Supplementary materials for this article are available online.
This study explores high school students' (n = 313) perceptions of and attitudes toward R programming in statistics and data science courses that use the CourseKata curriculum, which integrates R coding within an interactive statistics textbook. Specifically, we explore (a) shifts in confidence, positive sentiment, anxiety, perceived difficulty, and enjoyment from the beginning to the end of the course, (b) the predictive relationships between students'attitudes toward R programming and their future interest in statistics and data science, and whether there are differential patterns by racially marginalized background (including Black, Latine, and other racially marginalized groups). Findings reveal positive shifts in students' confidence and sentiment, coupled with reductions in anxiety and perceived difficulty. For perceived difficulty, racially marginalized students showed less pronounced declines over time compared to non-marginalized students. Predictive analyses highlight that post-course confidence and enjoyment in learning R strongly predict future interest in statistics, particularly for racially marginalized students. These findings underscore the potential of integrated R programming curricula to foster positive attitudes toward coding, while also emphasizing the need to address equity gaps in students'experiences.
