The Development of Expertise in Data Analysis Skills: An Exploration of the Cognitive and Metacognitive Processes by which Scientists and Students Construct Graphs

In the sciences, proficiency in graphing is held as a central element for literacy given the importance of succinctly communicating complex information. The need for graph literacy is even more evident when considering that television and the Internet commonly present data representations that inform opinion on public policy and personal actions. However, despite this recognition there is little understanding of how students develop the skills to interpret and represent data. As part of a larger mixed-methods project to investigate differences in data analysis skills along a continuum of expertise – from novice undergraduates to practicing science professionals, the purpose of this analysis is to understand the processes by which scientists and science students transform data into visual representations. Participants (n=32) across a spectrum of science expertise (i.e. science and non-science: faculty, graduate students, and undergraduates) generated graphs from four sets of tabular data from popular science textbooks while “thinking aloud” to expose their cognitive and metacognitive processes during task completion. Data were collected using a SMARTBoard system that captured participants’ pen strokes, and synced these with the audio recording to provide a temporal context. Visual data representations and audio recordings were assessed through content analysis using criteria focused on graph design, data relationships, and metacognitive practices. Regardless of level of expertise, participants were found to hold general deficiencies in graph skills (i.e. data pairing, inclusion of technical elements, and appropriate graph selection for the data type). While the results indicated a general positive trend between visual graphing skills and educational background, preliminary examinations of the “think aloud” recordings revealed substantial differences in the cognitive and metacognitive processes of participants along the expertise continuum in graph construction. Experts were more likely to reference the role of data type and variable relationships in guiding graph construction, while novices commonly misused graphing terminology and cited aspects such as ease, “look”, and personal preference in graph design. The observed general patterns begin to offer greater definition of the relationship between skill development and science expertise and to outline common graphing difficulties demonstrated by students. Results from this work provides for a deeper understanding of the development of data analysis skills to delineate where students fit along this progression. Instructionally, the findings identify certain areas where hang-ups in reading and creating graphs are likely to occur.