Informing and Evaluating Educational Applications With the Kirkpatrick Model in Virtual Environments: Using a Virtual Human Scenario to Measure Communication Skills Behavior Change

Increasingly, virtual environments are being used in educational and training applications. As with other types of applications that use virtual environments, these scenarios must be evaluated in terms of user experience. However, they also should be evaluated on the efficacy of the training or learning provided, so as to ensure learning transfer. Frameworks, such as the Kirkpatrick Model, exist to evaluate training scenarios, but application of these frameworks has not been fully utilized in development of virtual environment-based education and training. To address this gap and to also share our process with other virtual environment developers, we discuss our experience applying the Kirkpatrick Model to an existing virtual human (VH) application for medical communication skills training. The Kirkpatrick Model provides different levels of evaluation for training programs that include learners’ reactions to the training, the knowledge acquired from the training, behaviors indicating the training was applied, and the degree high-level results were impacted as a result of the training. While we discuss all of the Model’s levels, our focus for this work is Level 3 Behavior. The Kirkpatrick Model currently recommends that behavioral change may only be measured while a trainee is working in a real-world context. However, given existing evidence that VH applications have been shown to elicit real-world behaviors from participants, we suggest that VH training scenarios may be a method of measuring Behavior level metrics before trainees are evaluated in situ. Initial support for this suggestion is provided by our study examining whether VHs can elicit changes in communication skills learners’ message production behavior over time. This study indicates that learners displayed changes in several metrics over the course of the semester. Based on this finding, we suggest a direction for future research: observing learner behavior in a virtual environment as a pre-cursor to behavioral measures while in a real-world scenario.