Statistics and Data Science Education
Research in this area studies methods and mechanisms for teaching, learning, and assessment in statistics and data sciences education including comprehensive preparation for aspiring professionals as well as basic tenets of statistical reasoning and data literacy to prepare the broader populace to critique information as we make decisions and shape our views.
Penn State’s Department of Statistics is home to a vibrant and growing community of statistics and data science education researchers. Dennis Pearl’s research has pioneered use of technology to individualize learning, providing effective resources and professional development opportunities to teachers, using fun materials in teaching, and understanding how students interact with software. For example, he and Neil Hatfield lead an ongoing program which engages undergraduate students in both programming web-based educational applets and then conducting classroom research investigating the educational benefit of such interventions. Matthew Beckman and Marjorie Bond are interested in research and development for direct and indirect assessment instruments measuring a range of purposes including cognitive processes, learning outcomes, program evaluation, self-efficacy, affect, motivation, and anxiety.
Penn State Department of Statistics has been home to the Consortium for the Advancement of Undergraduate Statistics Education (CAUSE) since 2014. CAUSE programs are utilized by approximately 1 in 5 statistics instructors nationally in collaboration with 74 institutional members. Its mission is to support the advancement of undergraduate statistics education, in four target areas: resources, professional development, outreach, and research. Dennis Pearl is co-Founder and Director of CAUSE and Matt Beckman is an Associate Director for Research, a role and title he shares with Laura Le at University of Minnesota.
Faculty and Student Research Collaborations
NLP-assisted formative assessment feedback
Graduate student Susan Lloyd worked with Matthew Beckman and Dennis Pearl, in collaboration with computational linguist Rebecca Passonneau and her students Zhaohui Li and Zekun Wang, to investigate NLP-assisted formative assessment feedback for short-answer tasks in large enrollment statistics classes.
Research suggests “write-to-learn” tasks improve learning outcomes, yet free-text methods of formative assessment (e.g., short-answer questions) become unwieldy for large class instructors to review and respond in a reliable or timely manner. This project evaluates natural language processing (NLP) algorithms to assist this aim. The study found substantial inter-rater agreement between the algorithm and human reviewers when initially scoring tasks (i.e., essentially correct, partially correct, incorrect) with accuracy similar to the agreement observed from one human rater to the next. With compelling rater agreement, the study then pilots cluster analysis of response text toward enabling instructors to ascribe meaning to clusters in the interest of scalable formative assessment using free-text tasks.
Lloyd, S.E., Beckman, M.D., Pearl, D.K., Passonneau, R.J., Li, Z., & Wang, Z. (in review). Foundations of NLP-assisted formative assessment feedback for short-answer tasks in large enrollment statistics classes. Preprint URL: http://arxiv.org/abs/2205.02829
Web-based Teaching Applets
Teams of undergraduate students led by Dennis Pearl and Neil Hatfield have published web-based educational applets in the Book of Apps for Statistics Teaching (BOAST). Student cohorts have been designing, writing and refining the applets using the RShiny platform since 2017, and then the applets are freely available for instructors around the world to use in the classroom. Some of the students extend the experience by conducting educational research using the BOAST tools they have developed leading to peer reviewed publication or conference presentations (e.g., Wang et al., 2021).
Wang, S.L., Zhang, A.Y., Messer, S., Wiesner, A., & Pearl, D.K. (2021). Student-developed shiny applications for teaching statistics. Journal of Statistics and Data Science Education, 29(3). URL: https://www.tandfonline.com/doi/full/10.1080/26939169.2021.1995545