Collaborative Modeling With Complex Public Data Sets in the Middle School Classroom

• In Event: 63.025 - Supporting Modeling Epistemologies in the Science Classroom

Abstract

Objectives. Understanding the relationship between models and data is an important component of modeling practice. While research has explored how constructing data can help students develop productive modeling epistemologies (Manz, 2016), less has focused on the role of second-hand data in this process. Evidence suggests students treat these data differently for a variety of reasons (Delen & Krajcik, 2015; Kerlin, McDonald, & Kelly, 2010), in ways that may affect how they approach modeling tasks. Given that using data collected by others is a major component of scientific practice (Duschl, 2008) and allows a breadth of investigations (Hug & McNeill, 2008), the research question driving this study is: Under what circumstances do students come to treat public datasets as a source of evidence (McNeill & Berland, 2017) for constructing, defending, or evaluating scientific models?

Theoretical Framing. Practice-oriented perspectives toward science education seek to engage learners in data analysis and modeling as ways of building knowledge about natural phenomena of interest (NRC, 2011). Complex, publicly-sourced datasets can allow students to come together as community of learners (Brown & Campione, 1994) to answer questions of interest by pursuing different, but interrelated, paths of inquiry that inform the co-construction of models over time (Author, In Press). We acknowledge the important role that representation plays in supporting students’ sensemaking about data, and are interested in how particular tools and representations may serve as “model types” for students, supporting certain “modeling strategies” (White, Collins, & Frederiksen, 2011) that may be especially productive for collaborative data model construction.

Methods/Data. With two 7th grade science educators who taught a total of nine classes (over 300 students) in public middle schools in Northern California, we enacted three week-long data-rich science units that featured data sourced from the National Parks Service and the USGS. Units focused on key phenomena including trophic cascades in Yellowstone, the recent California water crisis, and patterns in earthquakes along major global fault lines. Students formulated questions, explored subsets of data using the CODAP tool (Finzer & Damelin, 2015), wrote arguments using a claims-evidence-reasoning format, and used those arguments to co-construct and revise models of the data and of the scientific systems that underlie them. Written work, video of classroom and group interactions, screen capture of data analysis, and observational field notes were collected from consented students.

Results and Scholarly Significance. Beginning with written work, we identified student groups that treated the available data as a source of evidence. Criteria for identification were that members of the group were chaining model elements using data, transforming data in order to investigate questions or build models, or challenging peers’ models using data as evidence. We analyzed video from these groups and found that flexibility in: (1) investigative path, (2) representations possible within the CODAP environment, and (3) ways of responding to peer contributions facilitated students’ use of data-as-evidence. This work reveals instructional and material supports for helping students approach publicly-sourced data as transformable evidence, rather than as merely authoritative resource, during modeling activities.

Authors

• Michelle Hoda Wilkerson, University of California - Berkeley

• Vasiliki Laina, University of California - Berkeley

• Lisette Lopez

• Rebecca L. Shareff, University of California - Berkeley

• Deniz Dogruer

• William McEachen