Multiple Representations in Science Learning and Assessment

• In Event: 35.030 - Multiple Representations and Multimedia: Student Learning and Instruction

Abstract

Purpose
Scientists rely on multiple forms of representation generate and communicate scientific theories and models (Bazerman, 1998; Goldman & Bisanz, 2002; Shanahan & Shanahan, 2008). Instruction that introduces students to, and engages them in, science practices necessarily involves learning from multiple representational forms including traditional text, drawings, diagrams, graphs, tables, and dynamic simulations (Achieve, 2013; CCSSO, 2010; Greenleaf, Brown, Goldman, & Ko, 2013; Magnusson & Palincsar, 1995; Wu & Krajcik, 2006). Likewise, assessments that are aligned with instruction need to employ multiple representational forms. This presentation examines estimates of learning from multiple representations derived from assessment tasks that varied in representational form, specifically essays and graphical displays, as well as production demands.

Significance
This work contributes to deepening the knowledge base regarding challenges and affordances of multiple representations in science learning and assessment. This is critical to ensuring that opportunities to learn science are accessible and implementable across diverse international populations.


Method
Adolescent students read a set of documents and answered a how/why question about the causes of coral bleaching (n=97) or skin cancer (n= 112). They were provided with the topic-appropriate document set that included multiple representation types (e.g., descriptive texts, images, graphs, and maps). Each document contributed only part of the overall causal model (including initiating factors, intervening variables, and the outcome). The only direct connections to the outcome included a link to an initiating factor in a graphical text and a link to an intervening variable in one verbal text. Students performed a series of assessment tasks with the documents available. The primary task was a high production, verbal task: write an essay explaining the causes of the phenomenon using information from the documents to support the conclusion. In a second task students selected the better of two graphical models of potential explanations and explain why it was best. Another lower production task included two multiple-choice questions: a verbal item asking students to pick the best explanation (testing key intervening causal elements) and a graph interpretation item.

Results and Conclusions
Overall preliminary results suggest that representational form in both the document set and the task format for assessment are impacting what students learn and the likelihood that they are able to make that learning visible. For example, for coral bleaching, only about 10% of the students wrote essays that connected information in the graph to textual information , although 27% mentioned the link between bleaching and the initiating cause. In contrast, for the graphical display task, 58% of the students correctly selected the model with intervening causal elements, but of those, only 62% were able to verbally justify this selection with reference to the causal chain. When intervening elements were tested in a verbal multiple choice format, 27% answered correctly, whereas 44% answered the graph interpretation question correctly. Analysis of the skin cancer topic is underway, and we are continuing to explore the individual patterns to better understand how the type of representation and the verbal production demands impact students’ ability to make their thinking visible.

Authors

• Anne Britt, Northern Illinois University

• Dylan Blaum, Northern Illinois University

• Patricia Wallace, Northern Illinois University

• Mon-Lin Ko, University of Illinois at Chicago

• Susan R. Goldman, University of Illinois at Chicago