Academic journal article Journal of Geoscience Education

How Students Reason about Visualizations from Large Professionally Collected Data Sets: A Study of Students Approaching the Threshold of Data Proficiency

Academic journal article Journal of Geoscience Education

How Students Reason about Visualizations from Large Professionally Collected Data Sets: A Study of Students Approaching the Threshold of Data Proficiency

Article excerpt

Introduction

Benchmarks for Science Literacy (American Association for the Advancement of Science, 2008), A Science Framework for K-12 Science Education (National Research Council [NRC], 2012), Next Generation Science Standards (NGSS Lead States, 2013), and many educators (Manduca & Mogk, 2002) emphasize students' direct engagement with data to develop scientific habits of mind and practices. Working with real data can improve students' reasoning about uncertainty and their quantitative skills (Creilson et al., 2008). Although scientists often reason about smallscale data sets from experiments and observations, reasoning about multiple and large-scale data sets has become increasingly important in science (National Research Council, 2010; Wolkovich, Regetz, & O'Connor, 2012). Having access to large-scale data sets allows scientific breakthroughs that could not be achieved from data collected by a single researcher, and may allow students and early career researchers to ask and answer bigger questions than would be possible if they were limited to only data they had collected themselves (Kastens, 2012; Linik, 2015; National Research Council, 2010).

To effectively incorporate large, professionally collected data into student activities, we need to better understand how such data is read, analyzed, and interpreted by students at different levels of mastery. However, most science education research on students' understanding of data involves small, student-collected data sets, and findings from such studies may not carry over to large, professionally collected datasets (Kastens, Krumhansl, & Baker, 2015). When students collect their own data, they develop an embodied understanding of the setting, methods, and potential data issues such as missing data and measurement error (Hug & McNeill, 2008), which they may lack with other-collected data sets. Interpretation of professionally collected data sets may also involve domain-specific knowledge of the referent systems and of specialized representational strategies.

Reasoning about large, professionally collected data can require the coordination of multiple sources of data (which can be from different times and locations), as well as an assortment of data-based visualizations and conceptual process models (which provide candidate explanations for the observed phenomena). Being able to connect different scientific ideas to form a coherent model is referred to as knowledge integration (e.g., Clark & Linn, 2003/2009; Linn, 2000). Scientific knowledge integration can be challenging for students, especially in geoscience, which draws from such a range of disciplines (Kastens & Manduca, 2012). An added difficulty is that na€ıve conceptions are often robust to change (Chi, 2005). Openness to conceptual change depends on the learner's level of engagement, depth and organization of background knowledge, motivation, disposition, willingness to engage with complex messages, and perception of the new content as understandable, coherent, plausible, and compelling (Dole & Sinatra, 1998; Lombardi, Sinatra, & Nussbaum, 2013). To change one's conceptual model when faced with data that disagree with the model is a sophisticated habit of mind that students often lack (Chi, 2005).

As described above, reasoning about data involves a host of different skills and knowledge. Consequently, developing a construct for how and how well people reason about data is challenging, because such reasoning is multifaceted (Mandinach & Gummer, 2012). Defining such an important construct is crucial, however, for developing appropriate teaching materials as well as valid and reliable assessments (Pellegrino, Wilson, Koenig, & Beatty, 2014). There has been a recent focus on defining data literacy versus data fluency (e.g., Greenberg & Walsh, 2012; Mandinach & Gummer, 2012; Manduca & Mogk, 2002). These terms imply empirically distinct stages (i.e., achieving literacy and fluency); however, conceptualizing one's ability to reason about data as a continuum is aligned with the learning progression framework presented by the National Research Council (2007, 2014). …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.