Concrete and Virtual Manipulative Models in Chemistry

• In Event: 63.031 - Connecting Virtual Tools and Physical Activities to Enhance Science Learning

Abstract

One reason that concrete models benefit learning is because they can be used to perform external actions that augment or replace difficult mental processes (Kirsh, 1995). Virtual models (that is, 3D computer visualizations) are replacing concrete models as computers become more powerful and available (Dede, 2009; Trelease, 2008). However, we know little about the relative effectiveness of virtual and concrete models. The purpose of our research was to directly compare the use of virtual and concrete models in a STEM domain, chemistry, for which models are an essential tool.

Perspective
Chemists use different diagrams and models to represent the structure of molecules. Dash-Wedge, Newman, and Fischer projections are three diagrams that must be mastered by organic chemistry students. In contrast with models, which can directly represent the 3D structure of molecules, diagrams represent the 3D structure of a molecule in the two dimensions of the printed page. These three diagrams follow different conventions to represent 3D structure and depict the molecule from different perspectives (see Fig. 1). Students often have difficulty relating these diagrams; however, providing concrete models leads to improved diagram translation performance (Stull, Hegarty, Dixon, & Stieff, 2012). Are virtual models also helpful?

Methods
In two studies, accuracy and efficiency in using virtual or concrete models was compared for naïve and chemistry students. Students manipulated each model to match the orientation and conformation of a molecule in a provided diagram. Comparing performance provides insight into the usability of these models in organic chemistry instruction.

Both studies followed a within-subjects design, with model type as the independent variable. Participants completed 24 orientation matching trials (counter-balancing two blocks of 12 with concrete or virtual models). Dependent measures were accuracy and response time.

Materials
The virtual models were delivered with a virtual modeling system that provided stereo-depth cues and a co-located, hand-held, direct manipulation interface. However, virtual models differed from concrete models in that the shape of the hand-held interface and the viewed object were not congruent. In addition, the virtual models had limited interactivity, which allowed students to only rotate the model around the central bond, whereas all bonds could be rotated on the concrete models.

Results
Matching accuracy was high and not significantly different between the concrete and virtual models for either group of students, which suggests that both models are functionally equivalent in our task. However, performance was more efficient when using the virtual models for both groups of students. This efficiency is likely because the virtual models constrained interactivity to task-relevant interactions. This efficiency advantage was evident when students used the virtual models in advance of the concrete models.

Significance of the work
Our findings demonstrate that computer-based virtual models can be at least as effective as concrete models in supporting understanding of spatial structures. Further, the interface used to manipulate virtual models can be designed to benefit the user by constraining the problem space. These findings inform instructional design suggesting that virtual models need not necessarily incorporate every aspect of the referent object.

Authors

• Andrew T. Stull, University of California - Santa Barbara

• Trevor Barrett, University of California - Santa Barbara

• Mary Hegarty, University of California - Santa Barbara