Tools for Learning and Doing Data Science at the K–12 Level

• In Event: Foundations of K–12 Data Science Education

Abstract

Objectives. Over the past few decades, there has been a proliferation of tools for teaching data analysis at the middle and high school levels (Hancock et al., 1992; Konold, 2007; McNamara, 2019). Understanding and using such tools are often cited as one of the more enabling–and difficult–aspects of teaching about data at any level. Issues of access, complexity, functionality, and format abound. While a few frameworks for systematically exploring the affordances and constraints of such tools exist, most explore only one or compare a few tools. This paper instead provides a thorough analysis of the existing landscape of tools being used, as well as complements the analysis with a literature review of the evidence base exploring student learning with these technologies.

Perspective. Our analysis leverages McNamara’s (2019) framework, which builds on the work of Biehler (1997; 2019) to identify 10 key attributes of modern statistical computing tools. The attributes include, for example, easy entry for novice users (#2), flexible plot creation (#5), and simple support for narrative, publishing, and reproducibility (#9).

Mode of inquiry. This paper will conduct a first-hand analysis of common educational data analysis software tools used with K-12 students. We will use the framework to analyze the process of completing the same exemplar analysis task with each software package (exploring a simple set of relationships that are available for analysis within a large, geographically-linked dataset derived from the U. S. Environmental Protection Agency). We complement this first-hand analysis with a review of empirical literature, especially comparative and design-based work, that examines the role of particular data tools in curriculum enactment and student learning (e.g., Lee & Delaney, 2022) or presents comparative analyses of tools (e.g., Koyuncu & Wilkerson, 2022).

Evidence. Analysis is ongoing. We will focus on a set of 7-10 software tools that are commonly used in K-12 data oriented lessons (e.g. R and RStudio, CODAP/Tuva, Pyret/Bootstrap). Our selection of tools was informed by the literature exploring teachers’ approaches to data (e.g. Rosenberg et al., 2022) as well as consultation with the field during the process of planning the NASEM workshop and eliciting examples from practice. The specific tools are further grouped into larger umbrella “genres” including Spreadsheets; Scripting Languages; and Visual Interfaces.

Results. Our preliminary analysis suggests some distinct, complementary functions of different specific tools and tool genres. For example, visual data analysis tools as a genre tend to emphasize evaluating general associations and open exploration over quantitative precision. At the same time, specific design features of some tools may further support specific skills (CODAP’s support of hierarchy), and other features can support thematic practices across genres (support for narrative documentation of process).

Significance. While curriculum and pedagogy are the most determinative factors in supporting student learning, specific tools can shape and constrain enactment. Understanding the purpose, benefits, and key features of different software packages can allow for the development of more conceptually aligned and accessible data science curriculum at the K-12 level.

Authors

• Daniel Pimentel, University of Alabama
  Presenting Author

• Michelle Hoda Wilkerson, University of California - Berkeley
  Non-Presenting Author

• Nicholas Horton
  Non-Presenting Author