A Meta-analysis of augmented reality programs for education and training

The application of augmented reality (AR) for education and training has grown dramatically in recent years, resulting in an expansive research domain within a relatively short amount of time. Two primary goals of the current article are to (a) summarize this literature by determining the overall effectiveness of AR programs relative to alternative comparisons and (b) assess the extent that AR program effectiveness is influenced by aspects of hardware, software, outcome, context, and methodology. A meta-analysis of over 250 studies supports that AR programs produce learning outcomes that are, on average, three-fifths of a standard deviation larger than alternative comparisons. Our results surprisingly show that AR programs using head-mounted displays produce significantly smaller effects than those using other output hardware (e.g., smartphones and tablets), and programs using image recognition are no more effective than those using alternative input methods (e.g., QR codes). We further find that most other aspects do not significantly influence observed program effectiveness; however, studies with younger participants produced significantly larger effects, and naturalistic studies produced significantly larger effects than laboratory studies. In our discussion, we utilize these findings to suggest promising theoretical perspectives for the study of AR, and we highlight methodological practices that can produce more accurate research moving forward. Thus, the current article summarizes research on AR education and training programs, identifies aspects that do and do not influence program efficacy, and provides several avenues for future research and practice.