Arts-based pedagogies are designed for reasoning about complex visual information. So often, science education is seen as learning a collection of facts or concepts. Teachers hesitate to use data or experiments that contain “messiness” because it might lead to ambiguity in discussing a particular science concept, and create inauthentic datasets to provide a “clean” illustration of a scientific principle (Kjelvik & Schultheis, 2019). This misses an important need and opportunity in science teaching. When the public views science as black-and-white, right-or-wrong, they struggle to reason through the complexity of real situations and the introduction of scientific uncertainty and variability can undermine confidence (Fischoff & Davis, 2011). To build a citizenry equipped to grapple with complex data, teaching with data requires shifting away from a view of data as representing right and wrong answers, and toward a view of data as a way in which to construct meaning and identify or test hypotheses about the world around us.

In this way, we see a parallel between teaching with data and teaching with art. Whether presented with a work of art or a complex data visualization, if a learner is told what it conveys or means, they have gained no ability to construct an understanding of the visual information. For this reason, over the past 30 years, art educators have developed approaches, grounded in constructivist learning theory, to facilitate sense-making and aesthetic development through interaction with complex visual information in paintings, drawings, sculpture, and other fine art forms. While these arts-based approaches vary, they all use facilitation that supports careful observation, scaffolded by questioning that pushes learners to construct their understanding, grounded in their observations, reflections, and discussion with peers. In Visual Thinking Strategies (VTS), for example: “The classroom teacher poses a sequence of open-ended questions about a series of carefully selected images of art works. Discussions unfold in which students puzzle and construct meaning about the art works. The learners are given a lot of ‘time on task,’ and have ample opportunity to build meaning one way and then another. They are also exposed to the thinking of their peers, which can accelerate shifts in their own thinking.” (Housen, 2002). In addition, this approach does not require the educator to be an art expert, as it focuses on joint exploration and meaning-making with the learners at the center.

This instructional approach has been found to support development of important critical thinking skills in art museums and classrooms, including observation, interpretation, comparison-making, and flexible thinking (Adams, Foutz, Luke, & Stein, 2006; Housen, 2002; Kisida, Bowen, & Greene, 2016; RK&A, 2018). Evidence has also shown promise that arts-based skill-development is transferable to other, non-art disciplines (e.g., Jasani & Saks, 2013; Naghshineh, et al., 2008). We think this type of open-ended, learner-centered, constructivist approach is ideally suited as a foundation for a new direction in teaching with complex scientific data visualizations in a way that will develop more “sticky” understanding of scientific concepts and build lasting abilities to observe, deconstruct, and reason about the data visualizations that are pervasive in our world.

Data literacy has cross-curricular importance, but educational systems are not built to support their development. Modern society is awash in data, as innovations in data collection and infrastructure technologies are making large, complex data more immediately and widely available (Börner, Bueckle, & Ginda, 2019; NASEM, 2018; ODI, 2014). There is incredible potential for these data resources to transform the way we learn about ourselves, our communities, and our environment, allowing learners to become independent explorers and discoverers in a complex world. However, evidence indicates that the fundamental skills needed to interpret and think critically about data is lacking in the general public (Goodman, et al., 2013), including among those who identify as interested in science (Börner, et al., 2016). A panel of professional data analysts, science educators, and mathematics educators has called for a “revolution in education, placing data literacy at its core, integrated throughout K-16 education” (ODI, 2016, p.14). In recent years, there has been substantial growth in efforts to study the teaching and learning of a variety of data literacy skills (see review in Lee & Wilkerson, 2018). One literature synthesis found data literacy positioned at the intersection of mathematical reasoning skills, computer/data science skills, and the authentic context of a particular discipline (Kjelvik & Schultheis, 2019), which gives it cross-curricular relevance, but may contribute to the sense that it lacks a clear disciplinary “home” in current structures (Finzer, 2013).

Data literacy with professional, scientific visualizations is an important and transferable skillset for all, but under-supported for many. While much has been learned about quantitative reasoning from the use of student-collected data and student-created visualizations, we argue that there is significant value in deeper study of how to support data literacy through interaction with professionally-collected scientific data. As others have argued (Baumer, 2015; Gould et al., 2014; Kastens et al., 2015; Kerlin et al., 2010; Kjelvik & Schultheis, 2019), the complex and “messy” data that result from scientific research can create a rich context for students to engage in scientific practices and improve their ability to think critically about scientific data and related phenomena. Geospatial data visualizations, in particular, are rapidly developing tools that scientists use to transform oceans of numbers and complex equations into visual representations that can support exploration of large-scale and complex systemic relationships, as well as issues of data limitations and multiple interpretations. While students are drawn to and engaged by these data representations, due to the familiarity of their underlying maps and striking visual features, our understanding of how to effectively use them in instruction is an emerging and under-studied area (Lee & Wilkerson, 2018; Radinsky, et al., 2014). Moreover, the ability to interpret, reason about, and question the complex data visualizations that are produced by scientists and the media is a needed lifelong skill (Börner, Bueckle, & Ginda, 2019).

Teachers need effective strategies to integrate data into instruction without becoming data experts. Recent shifts toward an integrated approach call for work with authentic data in science education (NGA & CCSSO, 2010; NGSS Lead States, 2013), and an analysis of the NGSS found that 46% of the Performance Expectations involve data skills, which are distributed across all grade bands (Kastens, 2015). The use of authentic data in classrooms has significant potential to transform the way science is taught and learned, but realizing this potential requires support that goes beyond providing access to data and visualization tools. Knowing what to do with the data can be a daunting task for educators (Kjelvik & Schultheis, 2019; Lee & Wilkerson, 2018), and conceptual and technical challenges encountered while working with data can create significant barriers to teaching and learning (Busey, Krumhansl, Mueller-Northcott, & Kochevar 2015; Krumhansl, Peach, Foster, Busey & Baker, 2012; Sickler & Cherry, 2013; Sickler, Hirsch, Madura, & Hoyle, 2017). For teaching with professionally-collected geospatial data, a recent review emphasized the need for strategies that help students focus attention within a complex visualization, and the importance of modeling “how to examine and inspect such data,” including the use of scaffolds that support students to construct their understanding (Lee & Wilerkson, 2018). However, professional development that targets teachers’ data-related knowledge and practice often focuses on their use of data (e.g., student-generated assessment data) to inform instruction (Mandinach & Gummer, 2016). While there is a growing body of research on mathematics teacher preparation and professional development to support the use of authentic data (e.g., Casey et al., 2020), there is a need for evidence-based professional learning models and communities of practice that can support science teachers’ understanding and use of professionally collected data in their curriculum and instruction.