To help students learn, instructors often use visuals (e.g., pie charts of fractions, ball-and-stick models of chemical molecules). We assume that these visuals help students learn; but they can be confusing if students do not know how the visuals show information. For example, ball-and-stick models show oxygen atoms as red spheres; does that mean that oxygen is red? (It does not.) The goal of this project was to help students learn with visuals. We developed an educational technology for undergraduate chemistry, Chem Tutor, that helps students learn how visuals show information. Within the Chem Tutor, students use interactive visuals to solve chemistry problems. While they interact with the visuals, students receive feedback from Chem Tutor about whether they manipulated the visual correctly (e.g., whether they constructed a ball-and-stick model that shows a given molecule correctly). The research project determined how the Chem Tutor can help students learn about visuals in a way that enhances their ability to learn from visuals about complex content knowledge. Specifically, the project focused on two representation skills: the ability to make sense of how particular visual features show particular concepts, and students’ fluency with visuals (akin to fluency in a language). To this end, we developed trainings for Chem Tutor tailored to support sense-making and fluency. We then conducted a series of experiments will investigate how best to combine trainings for sense-making and fluency, with the goal to optimally enhance students’ learning of chemistry content knowledge.

A first step of the project was to develop the Chem Tutor system. To this end, the project team worked closely with educators at 2-year and 4-year colleges. We also tested prototypes of Chem Tutor activities with undergraduate students. Throughout the project, we used data from the experiments to iteratively improve the usability of the Chem Tutor activities.

The second step was to test in which sequence students should receive trainings for sense-making and fluency: should students receive sense-making trainings first or fluency trainings first? Two experiments tested different ways to sequence these trainings. Results showed that students’ prior chemistry knowledge suggested which training they should receive first. In other words, different students have different needs when it comes to the order in which they should receive sense-making trainings and fluency trainings.

The third step was therefore to formally test the effectiveness of different training schedules for students with different levels of prior knowledge. A set of three experiments tested different ways of combining sense-making and fluency trainings for students with low, medium, and high prior knowledge. The results suggested that students with low prior knowledge should receive only sense-making trainings, students with medium prior knowledge should receive a combination of both trainings with sense-making before fluency trainings, and students with high prior knowledge should receive fluency trainings before sense-making trainings. These findings suggest a trajectory of different sequences of these trainings as students gradually progress from low to medium to high knowledge levels over the course of a longer learning period.

The fourth step was therefore to formally test whether adapting the sequence of sense-making and fluency trainings based on students’ given knowledge level. Chem Tutor has the capability to detect a student’s current knowledge level. This capability is based on artificial intelligence: similar to how Netflix keeps track of a viewer’s preference for certain types of movies, Chem Tutor keeps track of what a student knows about the visuals. We developed an adaptive version of Chem Tutor that uses this capability to determine whether a student should receive sense-making trainings or fluency trainings. A final experiment tested the effectiveness of this adaptive version by comparing it to a static version that provided sense-making trainings followed by fluency trainings for all students. The two versions were tested in an experiment during an entire semester in an undergraduate chemistry course. The results showed that students who worked with the adaptive version of Chem Tutor had significantly higher learning gains of chemistry knowledge across the semester.

In sum, the project yielded an educational technology that adapts to the individual student’s learning progress. Thereby, it makes efficient use of instructional time. Chem Tutor is available for free and has been used to support chemistry learning at 2-year and 4-year colleges. The findings about how to support learning with visuals can inform the development of educational technologies for other disciplines.

Last Modified: 12/03/2020
Modified by: Martina A Rau