Usability of concrete and virtual models in chemistry instruction Andrew T. Stull ⇑ , Trevor Barrett, Mary Hegarty University of California, Department of Psychological & Brain Sciences, Santa Barbara, CA 93106-9660, USA article info Article history: Available online 13 July 2013 Keywords: Diagrammatic reasoning Representational competence Manipulatives Molecular models Organic chemistry Virtual reality abstract Virtual models are increasingly common in the modern science classroom, however little is known about the relative effectiveness of virtual and concrete models. We developed a virtual modeling system and tested the benefit of haptic cue fidelity, controlling for many other perceptual differences between con- crete and virtual models. In two studies we directly compared performance of students using this virtual model and using concrete models for tasks in the domain of organic chemistry. Students used either vir- tual or concrete models to match diagrams of molecules or compare the structures of molecules repre- sented by models and diagrams. The results indicated similar levels of accuracy and similar ratings of usability for virtual and concrete models but no effect of haptic cue fidelity. Greater efficiency with virtual models was observed when students matched diagrams and models, and this efficiency transferred to later use of concrete models. The efficiency benefit is attributed to interactive constraints of the hand- held interface to the virtual model, which helped students identify task-relevant information in the model and limited them to performing the most task-relevant interactions with the models. Ó 2013 Elsevier Ltd. All rights reserved. 1. Introduction Concrete models are common learning aids in many science, technology, engineering, and mathematics (STEM) disciplines, including elementary mathematics (e.g., McNeil & Uttal, 2009; Sowell, 1989; Threadgil-Sowder & Juilfs, 1980; Uttal, Scudder, & DeLoache, 1997), geology (Steer, Knight, Owens, & McConnell, 2005), physics (Zacharia & Olympiou, 2011), chemistry (Copolo & Hounshell, 1995; Gupta & Parkash, 1999; Stull, Hegarty, Dixon, & Stieff, 2012) and engineering design (Klahr, Triona, & Williams, 2007). Increasingly, virtual models (that is, 3D computer visualiza- tions) are replacing concrete models as computers and mobile digital devices become more powerful and available in science classrooms (Dede, 2009; Limniou, Roberts, & Papadopoulos, 2008; Reimer & Moyer, 2005; Trelease, 2008). However, we know little about the relative effectiveness of virtual and concrete models for learning. The purpose of this study was to directly com- pare the use of virtual and concrete models in a STEM domain, chemistry, for which models are an essential tool used in both teaching and research. A secondary purpose was to develop and test a virtual modeling system that would support future studies of the perceptual factors and cognitive conditions that contribute to effective design of virtual models for instruction. 1.1. Medium vs. method Comparing virtual and concrete models for learning brings to mind the long standing debate over the primacy of the medium of delivery vs. the method of instruction for promoting learning. Clark (1983), Clark (1994) has argued that the media of instruction (e.g., static or animated diagrams, concrete or virtual models) are just vehicles for delivery of instruction, and it is instructional methods (e.g., direct instruction vs. discovery learning) that make a difference to learning. In contrast, Kozma (1994a), Kozma (1994b) argued that certain media offer unique attributes and affordances that promote learning (e.g., an animation allows for di- rect perception of motion but an illustration requires effortful mental animation). The developing consensus is that media and method are complementary (Hastings & Tracey, 2004), however a debate remains within the teaching community because much va- lue is attributed to new-media technologies for promoting mean- ingful learning (Barnea & Dori, 2000; Nicholson, Chalk, Funnell, & Daniel, 2006; Zacharia & Olympiou, 2011) and many studies still confound media with method (Triona & Klahr, 2003). For the pur- pose of our studies, we define method as the practice employed to elicit a learning outcome and medium as the material used to implement the method. Recent studies evaluating the use of concrete and virtual mod- els, while controlling for method of instruction, question the importance of the medium of instruction. In a study that compared virtual and concrete models for teaching experimental design prin- ciples to young children, and controlled all other aspects of the instructional setting, Triona and Klahr (2003) demonstrated 0747-5632/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.chb.2013.06.012 ⇑ Corresponding author. Tel.: +1 805 636 9769. E-mail addresses: Andrew.Stull@psych.ucsb.edu (A.T. Stull), Trevor.Barrett@ psych.ucsb.edu (T. Barrett), Mary.Hegarty@psych.ucsb.edu (M. Hegarty). Computers in Human Behavior 29 (2013) 2546–2556 Contents lists available at SciVerse ScienceDirect Computers in Human Behavior journal homepage: www.elsevier.com/locate/comphumbeh