Leveraging mathematics software data to understand student learning and motivation during the COVID-19 pandemic

Abstract

School closures during the COVID-19 pandemic presented a threat to student learning and motivation. Suspension of achievement testing created a barrier to understanding the extent of this threat. Leveraging data from a mathematics learning software as a substitute assessment, we found that students had lower engagement with the software during the pandemic, but students who did engage had increased performance. Students also experienced changes in motivation: lowered mathematics expectancy, but also lower emotional cost for mathematics. Results illustrate the potential and pitfalls of using educational technology data in lieu of traditional assessments and draw attention to access and motivation during at-home schooling.

Keywords:

• Emergency distance learning
• mathematics education
• motivation
• learning technologies

Notes on contributors

Teomara (Teya) Rutherford is a faculty member in the Learning Sciences program within the University of Delaware School of Education. She holds a PhD in Learning, Cognition, and Development from the University of California, Irvine and a Juris Doctor from Boston University School of Law. Her research focuses on learning and motivation in digital environments. Twitter: @DrTeyaR

Kerry Duck is a postdoctoral researcher at the University of Delaware. He completed his Ph.D. in Educational Psychology at the University of Northern Colorado. His research interests include student and teacher motivation within STEM disciplines. Additionally, he has methodological interests within the use of Ecological Momentary Assessments/Interventions to facilitate student motivation and learning. ORCID id:https://orcid.org/0000-0002-3602-5205 Twitter: @DrKerryDuck Linkedin: www.linkedin.com/in/KerryDuckPhD

Joshua M. Rosenberg (PhD, Michigan State University) is an assistant professor of STEM education and faculty fellow at the Center for Enhancing Education in Mathematics and Sciences at the University of Tennessee, Knoxville. His research focuses on how learners think of and with data, particularly in science education settings. Twitter: @JRosenberg6432

Ray Patt is a doctoral student at the University of Delaware. He received his B.S. in Neuroscience and Linguistics from The Ohio State University. He is broadly interested in how children learn in digital environments, and specifically interested in the role language plays in learning.Twitter: @RayPatt

Disclosure statement

No potential conflcit of interest was reported by the authors.

Data availability statement

Those interested in the data associated with this research should contact Teomara Rutherford at [email protected].

Notes

1 pseudonym

2 For the average level outcome, the estimation did not converge using the default optimizer. We chose to use the optimx package optimizer (Nash, 2014) for this outcome, which led to the estimation converging (and yielded identical effects as those from when we used the default optimizer).

3 We calculated the Boneferroni-adjusted p-value by dividing the conventional alpha value of .05 by the number of outcomes (10) to arrive at .005.

Additional information

Funding

This work was supported by the National Science Foundation under Grant 1544273 and Grant 1845584.